1. Preface

1.1. Who Should Use This Guide

The EXPRESSCLUSTER X Maintenance Guide describes maintenance-related information, intended for administrators. See this guide for information required for operating the cluster.

1.2. How This Guide is Organized

1.3. EXPRESSCLUSTER X Documentation Set

The EXPRESSCLUSTER manuals consist of the following five guides. The title and purpose of each guide is described below.

EXPRESSCLUSTER X Getting Started Guide

This guide is intended for all users. The guide covers topics such as product overview, system requirements, and known problems.

EXPRESSCLUSTER X Installation and Configuration Guide

This guide is intended for system engineers and administrators who want to build, operate, and maintain a cluster system. Instructions for designing, installing, and configuring a cluster system with EXPRESSCLUSTER are covered in this guide.

EXPRESSCLUSTER X Reference Guide

This guide is intended for system administrators. The guide covers topics such as how to operate EXPRESSCLUSTER, function of each module and troubleshooting. The guide is supplement to the Installation and Configuration Guide.

EXPRESSCLUSTER X Maintenance Guide

This guide is intended for administrators and for system administrators who want to build, operate, and maintain EXPRESSCLUSTER-based cluster systems. The guide describes maintenance-related topics for EXPRESSCLUSTER.

EXPRESSCLUSTER X Hardware Feature Guide

This guide is intended for administrators and for system engineers who want to build EXPRESSCLUSTER-based cluster systems. The guide describes features to work with specific hardware, serving as a supplement to the Installation and Configuration Guide.

1.4. Conventions

In this guide, Note, Important, See also are used as follows:

Note

Used when the information given is important, but not related to the data loss and damage to the system and machine.

Important

Used when the information given is necessary to avoid the data loss and damage to the system and machine.

See also

Used to describe the location of the information given at the reference destination.

The following conventions are used in this guide.

Convention

Usage

Example
Bold
Indicates graphical objects, such as fields, list boxes, menu selections, buttons, labels, icons, etc.
In User Name, type your name.
On the File menu, click Open Database.

Angled bracket within the command line

Indicates that the value specified inside of the angled bracket can be omitted.

clpstat -s[-h host_name]

#

Prompt to indicate that a Linux user has logged on as root user.

# clpcl -s -a

Monospace

Indicates path names, commands, system output (message, prompt, etc.), directory, file names, functions and parameters.

/Linux/5.2/en/server/
bold
Indicates the value that a user actually enters from a command line.
Enter the following:
# clpcl -s -a
italic
Indicates that users should replace italicized part with values that they are actually working with.
rpm -i expresscls-<version_number> -<release_number>.x86_64.rpm

EXPRESSCLUSTER X In the figures of this guide, this icon represents EXPRESSCLUSTER.

1.5. Contacting NEC

For the latest product information, visit our website below:

https://www.nec.com/global/prod/expresscluster/

2. The system maintenance information

This chapter provides information you need for maintenance of your EXPRESSCLUSTER system. Resources to be managed are described in detail.

This chapter covers:

2.1. Directory structure of EXPRESSCLUSTER

Note

Executable files and script files that are not described in "EXPRESSCLUSTER command reference" in the "Reference Guide" can be found under the installation directory. Run these files only with EXPRESSCLUSTER. Any failure or trouble caused by executing them by applications other than EXPRESSCLUSTER is not supported.

EXPRESSCLUSTER directories are structured as described below:

List of directories for EXPRESSCLUSTER installed

Fig. 2.1 Directory structure

  1. Directory for alert synchronization
    This directory stores EXPRESSCLUSTER Alert Synchronization's modules and management files.
  2. Directory for cluster modules
    This directory stores the EXPRESSCLUSTER Server's executable files.
  3. Directory for cloud environment
    This directory stores script files for cloud environment.
  4. Directory for cluster drivers
    • Mirror driver
      This directory stores the executable files of the data mirror driver.
    • Kernel mode LAN heartbeat, keepalive driver
      This directory stores the executable files of the kernel mode LAN heartbeat and keepalive driver.
  5. Directory for cluster configuration data
    This directory stores the cluster configuration files and policy file of each module.
  6. Directory for HA products linkage
    This directory stores binaries and configuration files for the Java Resource Agent and System Resource Agent.
  7. Directory for cluster libraries
    This directory stores the EXPRESSCLUSTER Server's library.
  8. Directory for licenses
    This directory stores licenses for licensed products.
  9. Directory for module logs
    This directory stores logs produced by each module.
  10. Directory for report messages (alert, syslog, mail)
    This directory stores alert, syslog and mail messages reported by each module.
  11. Directory for mirror disk and hybrid disk
    This directory stores the executable files and policy files etc. of the modules for mirror disk and hybrid disk.
  12. Directory for the performance logs
    This directory stores the information of performance about disk and system.
  13. Directory for EXEC resource script of group resources
    This directory stores EXEC resource scripts of group resources.
  14. Directory for the recovery script
    This directory stores the script executed by this function when an error is detected in the monitor resource if execution of a recovery script is in effect.
  15. Directory for temporary files
    This directory stores archive files created when logs are collected.
  16. Directory for the WebManager server and Cluster WebUI.
    This directory stores the WebManager's server modules and management files.
  17. Directory for module tasks
    This is a work directory for modules.
  18. usr/lib64
    This directory stores the symbolic links to the EXPRESSCLUSTER Server's library.
  19. /usr/sbin
    This directory stores the symbolic links to the EXPRESSCLUSTER Server's executable files.
  20. /etc/init.d
    For init.d environment, this directory stores the EXPRESSCLUSTER Service's Start/Stop scripts.
  21. /lib/systemd/system (for SUSE Linux, the path will be /usr/lib/ systemd/system.)
    For systemd environment, the setting file of EXPRESSCLUSTER service is stored in this directory.

2.2. How to delete EXPRESSCLUSTER logs or alerts

To delete EXPRESSCLUSTER logs or alerts, perform the following procedure.

  1. Disable all cluster services on all servers in a cluster .

    clpsvcctrl.sh --disable -a

  2. Shut down the cluster with the Cluster WebUI or clpstdn command, and then reboot the cluster.

  3. To delete logs, delete the files and directories in the following directory. Perform this operation on the server for which you want to delete the logs.

    • /opt/nec/clusterpro/log/

  4. To delete alerts, delete the files in the following directory. Perform this operation on the server for which you want to delete the alerts.

    • /opt/nec/clusterpro/alert/log/

  5. Enable all cluster services on all servers in a cluster .

    clpsvcctrl.sh --enable -a

  6. Run the reboot command on all the servers in the cluster to reboot the cluster.

2.3. Mirror statistics information collection function

If the Mirror Statistics check box is already checked on the Statistics tab of Cluster Properties in the config mode of Cluster WebUI, information on the mirror performance is collected and saved to <installation path>/perf/disk according to the following file naming rules. In the following explanations, this file is represented as the mirror statistics information file.

nmpN.cur
nmpN.pre[X]

cur

Indicates the latest information output destination.

pre

Indicates the previous, rotated, information output destination.

N

Indicates the target NMP number.

[X]
Indicates the generation number.
For a file that is one generation older, the generation number is omitted.
For a file that is m generations older, X is assumed to be m-1.
If the total number of generations is n, X of the oldest file is assumed to be n-2.

The collected information is saved to the mirror statistics information file. The time during which statistics information is output to this file (=sampling interval) is 60 seconds. .If the size of current log file reached 16MB, it is rotated to new log file. And two generation log files can be saved. Information recorded to the mirror statistics information file can be used as a reference for the tuning related to the mirror function. The collected statistics information contains the following items.

Note

The extracted mirror statistics information is included in the logs collected by the clplogcc command or Cluster WebUI.
Specify type5 to collect the log by the clplogcc command; specify Pattern 5 to collect the log by the Cluster WebUI. For details about log collection, see "Collecting logs (clplogcc command)" in "EXPRESSCLUSTER command reference" in the Reference Guide or the online manual.
Statistic value name
Unit
1
Description
Output
2
Write, Total
(Write amount)
Byte
(MB)
Total amount of data written to the mirror partition
The value to be output is the amount of data written by every sampling.
LOG,
CMD
(A)
Write, Avg
(Write amount, average value)
Byte/s
(MB/s)
Amount of data written to the mirror partition per unit time
LOG,
CMD
(A)
Read, Total
(Read amount)
Byte
(MB)
Total amount of data read from the mirror partition
The value to be output is the amount of data read by every sampling.
LOG,
CMD
(A)
Read, Avg
(Read amount, average value)
Byte/s
(MB/s)
Amount of data read from the mirror partition per unit time
LOG,
CMD
(A)
Local Disk Write, Total
(Local disk write amount)
Byte
Total amount of data written to the local disk (data partition)
The value to be output is the amount of data written by every sampling.
LOG
(B)
Local Disk Write, Avg
(Local disk average write amount)
Byte/s
Amount of data written to the local disk (data partition) per unit time
LOG
(B)
Local Disk Read, Total
(Local disk read amount)
Byte
Total amount of data read from the local disk (data partition)
The value to be output is the amount of data read by every sampling.
LOG
(B)
Local Disk Read, Avg
(Local disk average read amount)
Byte/s
Amount of data read from the local disk (data partition) per unit time
LOG
(B)
Send, Total
(Mirror communication amount, total value)
Byte
(KB)
Total amount of mirror communication sent up until mirror disk connect

The value to be output is the communication amount by every sampling.
TCP control information and the like are excluded.
LOG,
CMD
(B)
Send, Avg
(Mirror communication amount, average value)
Byte/s
(KB/s)
Mirror communication amount sent by up until mirror disk connect per unit time
LOG,
CMD
(B)
Compress Ratio
(Compression ratio)
%
Mirror data compression ratio
(Post-compression size) / (pre-compression size)
x 100

100 for noncompression
The value to be output is calculated based on the communication data for every sampling.
LOG
(A)
Sync Time, Max
(Mirror communication time, maximum value)
Second/time
Time needed until the first piece of mirror synchronization data is synchronized.[#f3]_ The value to be output is the longest mirror synchronization data time.

Mirror synchronization data that failed to be synchronized due to non-communication or the like (resulting in a mirror break) is excluded.
Moreover, the value to be output is obtained for communication for every sampling.
LOG,
CMD
(A)
Sync Time, Avg
(Mirror communication time, average value)
Second/time
Time needed until the first piece of mirror synchronization data is synchronized. 3 The value to be output is the average for all the communications.

Mirror synchronization data that failed to be synchronized due to non-communication or the like (resulting in a mirror break) is excluded.
Moreover, the value to be output is obtained for communication for every sampling.
LOG,
CMD
(A)
Sync Ack Time, Max
(Mirror synchronization ACK response time, maximum value)
Millisecond
Time that elapses between mirror synchronization data being sent to the other server and ACK being received from the other server. 3 The maximum value of all such times is output.

This value is used as a reference to determine Ack Timeout of the Mirror Driver tab that is set with the mirror disk resource or hybrid disk resource.

However, mirror synchronization data that results in an ACK timeout is excluded from the measurement.
The value to be output is the time after the mirror daemon (mirror agent) starts.
LOG
(A)
Sync Ack Time, Cur
(Mirror synchronization ACK response time, latest value)
Millisecond
Of the lengths of time needed for mirror synchronization data ACK reception, this value is the time that needed for the most recent ACK reception. 3

However, mirror synchronization data that results in an ACK timeout is excluded from the measurement.
LOG
(A)
Recovery Ack Time, Max
(Mirror recovery ACK response time, maximum value)
Millisecond
Time that elapses between mirror recovery data being sent to the other server and ACK being received from the other server
The maximum value of all such times is output.

This value is used as a reference to determine Ack Timeout of the Mirror Driver tab that is set with the mirror disk resource or hybrid disk resource.

However, mirror synchronization data that results in an ACK timeout is excluded from the measurement.
The value to be output is the time after the mirror daemon (mirror agent) starts.
LOG
(A)
Recovery Ack Time, Max2
(Mirror recovery ACK response time, maximum value during a certain period)
Millisecond
Maximum value of the time that elapses between mirror recovery data being sent to the other server and ACK being received from the other server.

The maximum value during one sampling period is output.

However, mirror synchronization data that results in an ACK timeout is excluded from the measurement.
LOG
(A)
Recovery Ack Time, Cur
(Mirror recovery ACK response time, latest value)
Millisecond
Time that elapses between the mirror recovery data being sent to the other server and ACK being received from the other server
The value to be output is the time needed for the most recent ACK reception.

However, mirror synchronization data that results in an ACK timeout is excluded from the measurement.
LOG
(A)
Sync Diff, Max
(Difference amount, maximum value)
Byte
(MB)
Amount of mirror synchronization data that has not yet been synchronized with the other server. The value to be output is the maximum from among all the samplings.

Mirror synchronization data that failed to be synchronized due to non-communication or the like (resulting in a mirror break) is excluded.
LOG,
CMD
(A)
Sync Diff, Cur
(Difference amount, latest value)
Byte
(MB)
Amount of mirror synchronization data that has not yet been synchronized with the other server. The value to be output is that which was used most recently for collection.

Mirror synchronization data that failed to be synchronized due to non-communication or the like (resulting in a mirror break) is excluded.
LOG,
CMD
(A)
Send Queue, Max
(Number of send queues, maximum value)
Quantity
Number of queues used when mirror synchronization data is sent. The value to be output is the maximum used after the mirror daemon (mirror agent) starts.

This value is used as a reference to determine Number of Queues in Asynchronous mode that is set with the mirror disk resource or hybrid disk resource.
LOG
(A)
Send Queue, Max2
(Number of send queues, maximum value during a certain period)
Quantity
Number of queues used when mirror synchronization data is sent. The maximum value during one sampling period is output.
LOG
(A)
Send Queue, Cur
(Number of send queues, latest value)
Quantity
Number of queues used when mirror synchronization data is sent. The value to be output is that which was used most recently for collection.
LOG
(A)
Request Queue, Max
(Number of request queues, maximum value)
Quantity
Number of I/O requests being processed that were sent to the mirror partition. The value to be output is the maximum used after the mirror daemon (mirror agent) starts.

This value is used as a reference to determine Request Queue Maximum Number of the Mirror Driver tab of cluster properties.
LOG
(A)
Request Queue, Max2
(Number of request queues, maximum value during a certain period)
Quantity
Number of I/O requests being processed that were sent to the mirror partition. The maximum value during one sampling period is output.
LOG
(A)
Request Queue, Cur
(Number of request queues, latest value)
Quantity
Number of I/O requests being processed that were sent to the mirror partition. The value to be output is that which was used most recently for collection.
LOG
(A)
MDC HB Time Max
(Mirror disconnect heartbeat time, maximum value)
Second
Time that elapses between ICMP ECHO being sent to the other server through mirror disconnect and ICMP ECHO REPLY being received from the other server.
The value to be output is the maximum used after the mirror daemon (mirror agent) starts.
LOG
(B)
MDC HB Time, Max2
(Mirror disconnect heartbeat time, maximum value during a certain period)
Second
Time that elapses between ICMP ECHO being sent to the other server through mirror disconnect and ICMP ECHO REPLY being received from the other server.
The maximum value during one sampling period is output.
LOG
(B)
MDC HB Time Cur
(Mirror disconnect heartbeat time, latest value)
Second
Time that elapses between ICMP ECHO being sent to the other server through mirror disconnect and ICMP ECHO REPLY being received from the other server.
The value to be output is that which was used most recently for collection.
LOG
(B)
Local-Write Waiting Recovery-Read Time, Total
(Mirror synchronization I/O exclusion time, total value)
Second
If writing to the same area of the disk occurs during mirror recovery, writing is held until the mirror recovery for that area is complete.
The value to be output is the cumulative value of the hold time, from when the mirror daemon (mirror agent) starts.

That hold time may be long if Recovery Data Size of the Mirror Agent tab of the cluster properties is made large. This value is used as a reference to determine this size.
LOG
(A)
Local-Write Waiting Recovery-Read Time, Total2
(Mirror synchronization I/O exclusion time, total value during a certain period)
Second
If writing to the same area of the disk occurs during mirror recovery, writing is held until the mirror recovery for that area is complete.
The value to be output is the cumulative value of the hold time during one sampling period.
LOG
(A)
Recovery-Read Waiting Local-Write Time, Total
(Mirror recovery I/O exclusion time, total value)
Second
If reading of mirror recovery data from the same area of the disk occurs during writing to the mirror partition, reading of the mirror recovery data is held until writing to that area is complete.
The value to be output is the cumulative value of the hold time, from when the mirror daemon (mirror agent) starts.

That hold time may be long if Recovery Data Size of the Mirror Agent tab of the cluster properties is made large. This value is used as a reference to determine this size.
LOG
(A)

Recovery-Read Waiting Local-Write Time, Total2

Second

If reading of mirror recovery data from the same area of the disk occurs during writing to the mirror partition, reading of the mirror recovery data is held until writing to that area is complete.

LOG

X(Mirror recovery I/O exclusion time, total value during a certain period)

The value to be output is the cumulative value of the hold time during one sampling period.
Unmount Time, Max
(Unmount time, maximum value)
Second
Time needed for unmount to be executed when the mirror disk resource or hybrid disk resource is deactivated

This value is used as a reference to determine Timeout of the Unmount tab that is set with the mirror disk resource or hybrid disk resource.
LOG
(A)
Unmount Time, Last
(Unmount time, latest value)
Second
Time needed for unmount to be executed when the mirror disk resource or hybrid disk resource is deactivated
The value to be output is the time needed when unmount was most recently executed.
LOG
(A)
Fsck Time, Max
(fsck time, maximum value)
Second
Time needed for fsck to be executed when the mirror disk resource or hybrid disk resource is activated

This value is used as a reference to determine fsck Timeout of the fsck tab that is set with the mirror disk resource or hybrid disk resource.
LOG
(A)
Fsck Time, Last
(fsck time, latest value)
Second
Time needed for fsck to be executed when the mirror disk resource or hybrid disk resource is activated
The value to be output is the time needed when fsck was most recently executed.
LOG
(A)
1
The unit in parentheses is used for command display. During output, a value of up to two decimal places is output. The third decimal place is truncated.
The conversion rules are as follows:
1 KB = 1024 bytes, 1 MB = 1048576 bytes
If a value is truncated to 0, "0.00" is output. If the value is 0 without truncation, "None" is displayed for commands, or "0" for the mirror statistics information file.
2
CMD : Information that is visible with commands (clpmdstat, clphdstat)
LOG : Information that is output to the mirror statistics information file
(A) : In case of Active, the valid value is output.
(B) : In both cases of Active/Standby, the valid value is output.
Further, only mirror statistics information on a local server is recorded, information on other servers is not recorded.
3(1,2,3)
If the mode is "synchronous", "time taken from sending a mirror synchronous data to receiving ACK from the other server".
If the mode is "asynchronous", "time taken from placing mirror synchronous data on the synchronization queue to receiving ACK from the other server".

Display with commands can be used only when Mirror Statistics is already enabled in the Statistics tab of Cluster Properties in Cluster WebUI.

2.4. System resource statistics information collection function

If the System Resource Statistics check box is already checked on the Statistics tab of Cluster Properties in the Cluster WebUI config mode and if system monitor resources or process resource monitor resources are already added to the cluster, information on the system resource is collected and saved under <installation path>/perf/system according to the following file naming rules.

This file is in CSV-format. In the following explanations, this file is represented as the system resource statistics information file.

system.cur
system.pre

cur

Indicates the latest information output destination.

pre

Indicates the previous, rotated, information output destination.

The collected information is saved to the system resource statistics information file. The time during which statistics information is output to this file (=sampling interval) is 60 seconds. .If the size of current log file reached 16MB, it is rotated to new log file. And two generation log files can be saved. Information recorded to the system resource statistics information file can be used as a reference for analyzing the system performance.The collected statistics information contains the following items.

Statistic value name

Unit

Description

CPUCount

Quantity

Number of CPUs

CPUUtilization

%

CPU utilization

CPUTotal

10 Millisecond

Total CPU time

CPUUser

10 Millisecond

CPU usage time in the user mode

CPUNice

10 Millisecond

CPU usage time in the user mode with low priority

CPUSystem

10 Millisecond

CPU usage time in the system mode

CPUIdle

10 Millisecond

CPU idle time

CPUIOWait

10 Millisecond

I/O wait time

CPUIntr

10 Millisecond

Interrupt processing time

CPUSoftIntr

10 Millisecond

Software interrupt processing time

CPUSteal

10 Millisecond

Time when CPU was consumed by the OS on another virtual machine for virtual environment

MemoryTotalSize

Byte (KB)

Total memory capacity

MemoryCurrentSize

Byte (KB)

Memory usage

MemoryBufSize

Byte (KB)

Buffer size

MemoryCached

Byte (KB)

Cache memory size

MemoryMemFree

Byte (KB)

Available memory capacity

MemoryDirty

Byte (KB)

Memory data waiting to be written on hard disk

MemoryActive(file)

Byte (KB)

Buffer or page cache memory

MemoryInactive(file)

Byte (KB)

Available buffer or available page cache memory

MemoryShmem

Byte (KB)

Shared memory size

SwapTotalSize

Byte (KB)

Available swap size

SwapCurrentSize

Byte (KB)

Currently used swap size

SwapIn

Times

Number of times of swap-in

SwapOut

Times

Number of times of swap-out

ThreadLimitSize

Quantity

Maximum number of threads

ThreadCurrentSize

Quantity

Current number of threads

FileLimitSize

Quantity

Maximum number of opened files

FileCurrentSize

Quantity

Current number of opened files

FileLimitinode

Quantity

Number of inodes in the whole system

FileCurrentinode

Quantity

Current number of inodes

ProcessCurrentCount

Quantity

Current total number of processings

The following output is an example of system resource statistics information file.

  • system.cur

    "Date","CPUCount","CPUUtilization","CPUTotal","CPUUser","CPUNice","CPUSystem","CPUIdle","CPUIOWait","CPUIntr","CPUSoftIntr","CPUSteal","MemoryTotalSize","MemoryCurrentSize","MemoryBufSize","MemoryCached","MemoryMemFree","MemoryDirty","MemoryActive(file)","MemoryInactive(file)","MemoryShmem","SwapTotalSize","SwapCurrentSize","SwapIn","SwapOut","ThreadLimitSize","ThreadCurrentSize","FileLimitSize","FileCurrentSize","FileLimitinode","FileCurrentinode","ProcessCurrentCount"
"2019/10/31 15:44:50","2","0","34607369","106953","59","23568","34383133","89785","0","3871","0","754236","231664","948","334736","186888","12","111320","167468","50688","839676","0","0","0","5725","183","71371","1696","22626","22219","121"
"2019/10/31 15:45:50","2","0","34619340","106987","59","23577","34395028","89816","0","3873","0","754236","231884","948","334744","186660","12","111320","167476","50688","839676","0","0","0","5725","183","71371","1696","22867","22460","121"
"2019/10/31 15:46:50","2","0","34631314","107022","59","23586","34406925","89846","0","3876","0","754236","231360","948","334764","187164","4","111348","167468","50688","839676","0","0","0","5725","183","71371","1696","22867","22460","121"
                                     :

2.5. Process resource statistics information collection function

If the System Resource Statistics check box is already checked on the Statistics tab of Cluster Properties in the Cluster WebUI config mode and if system monitor resources or process resource monitor resources are already added to the cluster, information on the process resource is collected and saved under <installation path>/perf/system according to the following file naming rules.

This file is in CSV-format. In the following explanations, this file is represented as the process resource statistics information file.

process.cur
process.pre

cur

Indicates the latest information output destination.

pre

Indicates the previous, rotated, information output destination.

The collected information is saved to the process resource statistics information file. The time during which statistics information is output to this file (=sampling interval) is 60 seconds. .If the size of current log file reached 32MB, it is rotated to new log file. And two generation log files can be saved. Information recorded to the process resource statistics information file can be used as a reference for analyzing the process performance.The collected statistics information contains the following items.

Statistic value name

Unit

Description

PID

-

Process ID

CPUUtilization

%

CPU utilization

MemoryPhysicalSize

Byte (KB)

Physical memory usage

MemoryVirtualSize

Byte (KB)

Virtual memory usage

ThreadCurrentCount

Quantity

Number of running threads

FileCurrentCount

Quantity

Number of opening files

ProcessName

-
Process name
* Outputted not in double quotes.

The following output is an example of process resource statistics information file.

  • process.cur

    "Date","PID","CPUUtilization","MemoryPhysicalSize","MemoryVirtualSize","ThreadCurrentCount","FileCurrentCount","ProcessName"
"2022/09/05 17:08:41","620","0","26384","1132","1","21",/usr/lib/systemd/systemd-logind
"2022/09/05 17:08:41","623","0","126384","1096","1","6",/usr/sbin/crond -n
"2022/09/05 17:08:41","1023","0","239924","2880","3","12",/usr/sbin/rsyslogd -n
                                     :

2.6. Cluster statistics information collection function

In the config mode of Cluster WebUI, with the Cluster Statistics check box (open Cluster Properties -> the Statistics tab) checked, CSV text files are created containing information on the processing results and time of, for example, reception interval for heartbeat resources, group failovers, starting group resources, and monitoring processes by monitor resources. These files are hereinafter called cluster statistics information files.

  • For heartbeat resources

    Information is outputted to the file for each heartbeat resource type. This function is supported by kernel mode LAN heartbeat resources and user mode LAN heartbeat resources.

    [Heartbeat resource type].cur
    [Heartbeat resource type].pre

    cur

    Indicates the latest information output destination.

    pre

    Indicates the previous, rotated, information output destination.

    File location

    <installation path>/perf/cluster/heartbeat/

  • For groups

    group.cur
    group.pre

    cur

    Indicates the latest information output destination.

    pre

    Indicates the previous, rotated, information output destination.

    File location

    <installation path>/perf/cluster/group/

  • For group resources

    The information for each type of group resource is output to the same file.

    [Group resource type].cur
    [Group resource type].pre

    cur

    Indicates the latest information output destination.

    pre

    Indicates the previous, rotated, information output destination.

    File location

    <installation path>/perf/cluster/group/

  • For monitor resources

    The information for each type of monitor resources is output to the same file.

    [Monitor resource type].cur
    [Monitor resource type].pre

    cur

    Indicates the latest information output destination.

    pre

    Indicates the previous, rotated, information output destination.

    File location

    <installation path>/perf/cluster/monitor/

Note

The cluster statistics information file is included in the logs collected by the clplogcc command or Cluster WebUI.

Specify type 6 to collect the log by the clplogcc command; specify Pattern 6 to collect the log by the Cluster WebUI. For details about log collection, see. "Collecting logs (clplogcc command)" in "EXPRESSCLUSTER command reference" of "Reference Guide" or the online manual.

Listed below are the timing to output the statistics information to the cluster statistics information file:

  • For heartbeat resources

  • Periodical output

  • For groups 4

  • When the group startup processing is completed

  • When the group stop processing is completed

  • When the group move processing is completed 5

  • When the failover processing is completed 5

  • For group resources

  • When the group resource startup processing is completed

  • When the group resource stop processing is completed

  • For monitor resources

  • When the monitor processing is completed

  • When the monitor status change processing is completed

4

If a single unit of group resource was started or stopped, the group statistics information is not output.

5(1,2)

If a group was moved or failed over, the statistics information is output to the failover target server.

The statistics information to be collected includes the following items:

  • For heartbeat resources

    Statistic value name

    Description

    Date
    Time when the statistics information is output.
    This is output in the form below (000 indicates millisecond):
    YYYY/MM/DD HH:MM:SS.000

    Name

    Name of a heartbeat resource.

    Type

    Type of the heartbeat resource.

    Local

    Host name of the local server.

    Remote

    Host name of the other server.

    RecvCount

    Heartbeat reception count during the log output interval.

    RecvError

    Error reception count during the log output interval.

    RecvTime(Min)

    Minimum interval (in milliseconds) of heartbeat reception during the log output interval.

    RecvTime(Max)

    Maximum interval (in milliseconds) of heartbeat reception during the log output interval.

    RecvTime(Avg)

    Average interval (in milliseconds) of heartbeat reception during the log output interval.

    SendCount

    Heartbeat transmission count during the log output interval.

    SendError

    Error transmission count during the log output interval.

    SendTime(Min)

    Minimum time (in milliseconds) for heartbeat transmission during the log output interval.

    SendTime(Max)

    Maximum time (in milliseconds) for heartbeat transmission during the log output interval.

    SendTime(Avg)

    Average time (in milliseconds) for heartbeat transmission during the log output interval.

  • For others (except heartbeat resources)

Statistic value name

Description

Date
Time when the statistics information is output.
This is output in the form below (000 indicates millisecond):
YYYY/MM/DD HH:MM:SS.000

Name

Name of group, group resource or monitor resource.

Action
Name of the executed processing.
The following strings are output:
For groups: Start (at start), Stop (at stop), Move (at move), Failover (at failover)
For group resources: Start (at activation), Stop (at deactivation)
For monitor resources: Monitor (at monitor execution)

Result
Name of the results of the executed processing.
The following strings are output:
When the processing was successful: Success (no errors detected in monitoring or activation/deactivation)
When the processing failed: Failure (errors detected in monitoring or activation/deactivation)
When a warning occurred: Warning (only for monitoring, in case of warning)
When a timeout occurred: Timeout (monitoring timeout)
When the processing was canceled: Cancel (canceling processings such as cluster shutdown during group startup)

ReturnCode

Return value of the executed processing.

StartTime
Start time of the executed processing.
This is output in the form below (000 indicates millisecond):
YYYY/MM/DD HH:MM:SS.000

EndTime
End time of the executed processing.
This is output in the form below (000 indicates millisecond):
YYYY/MM/DD HH:MM:SS.000

ResponseTime(ms)
Time taken for executing the processing (in millisecond).
This is output in millisecond.

Here is an example of the statistics information file to be output when a group with the following configuration is started up:

  • Server - Host name: server1, server2

  • Heartbeat resource

    • Kernel mode LAN heartbeat resource
      Resource name: lankhb1, lankhb2

  • Group

    • Group name: failoverA

  • Group resource which belongs to the group (failoverA)

    • exec resource
      Resource name: exec01, exec02, exec03

  • lankhb.cur

    "Date","Name","Type","Local","Remote","RecvCount","RecvError","RecvTime(Min)","RecvTime(Max)","RecvTime(Avg)","SendCount","SendError","SendTime(Min)","SendTime(Max)","SendTime(Avg)"
"2018/12/18 09:35:36.237","lankhb1","lankhb","server1","server1","20","0","3000","3000","3000","20","0","0","0","0"
"2018/12/18 09:35:36.237","lankhb1","lankhb","server1","server2","20","0","3000","3000","3000","20","0","0","0","0"
"2018/12/18 09:35:36.237","lankhb2","lankhb","server1","server1","20","0","3000","3000","3000","20","0","0","0","0"
"2018/12/18 09:35:36.237","lankhb2","lankhb","server1","server2","20","0","3000","3000","3000","20","0","0","0","0"
                                :

  • group.cur

    "Date","Name","Action","Result","ReturnCode","StartTime","EndTime","ResponseTime(ms)"
"2018/12/19 09:44:16.925","failoverA","Start","Success",,"2018/12/19 09:44:09.785","2018/12/19 09:44:16.925","7140"
                               :

  • exec.cur

    "Date","Name","Action","Result","ReturnCode","StartTime","EndTime","ResponseTime(ms)"
"2018/12/19 09:44:14.845","exec01","Start","Success",,"2018/12/19 09:44:09.807","2018/12/19 09:44:14.845","5040"
"2018/12/19 09:44:15.877","exec02","Start","Success",,"2018/12/19 09:44:14.847","2018/12/19 09:44:15.877","1030"
"2018/12/19 09:44:16.920","exec03","Start","Success",,"2018/12/19 09:44:15.880","2018/12/19 09:44:16.920","1040"
                                     :

2.6.1. Notes on the size of the cluster statistics information file

The number of cluster statistics information files to be generated differs depending on their configurations. Some configurations may cause a large number of files to be generated. Therefore, consider setting the size of the cluster statistics information file according to the configuration. The maximum size of the cluster statistics information file is calculated with the following formula:

The size of the cluster statistics information file =
([Heartbeat resource file size] x [number of types of heartbeat resources which are set]) x (number of generations (2)) +
([Group file size]) x (number of generations (2)) +
([Group resource file size] x [number of types of group resources which are set]) x (number of generations (2)) +
([Monitor resource file size] x [number of types of monitor resources which are set]) x (number of generations (2))

Example: For the following configuration, the total maximum size of the cluster statistics information files to be saved is 332 MB with this calculation. ((((50MB) x 1) x 2) + (((1MB) x 2) + ((3MB x 5) x 2) + ((10MB x 10) x 2) = 332MB)

  • Number of heartbeat resource types: 1 (file size: 50 MB)

  • Group (file size: 1 MB)

  • Number of group resource types: 5 (file size: 3 MB)

  • Number of monitor resource types: 10 (file size: 10 MB)

2.7. Function for outputting the operation log of Cluster WebUI

If the Output Cluster WebUI Operation Log check box is already checked on the WebManager tab of Cluster Properties in the config mode of Cluster WebUI, the information on the operation of Cluster WebUI is outputted to the log file. This file is in CSV format, which is hereinafter called "the operation log file of Cluster WebUI.

webuiope.cur
webuiope.pre<x>

cur

Indicates the last outputted log file.

pre<x>
Indicates a previously outputted but rotated log file.
pre, pre1, pre2, ..., in reverse chronological order.
When the prescribed number of existing log files is exceeded, the oldest log file is deleted.

Where to save

Directory as Log output path in the config mode of Cluster WebUI

The operation information to be outputted includes the following items:

Item name

Description

Date
Time when the operation information is outputted.
This is outputted in the form below (000 in milliseconds):
YYYY/MM/DD HH:MM:SS.000

Operation

Name of the executed operation in Cluster WebUI.

Request

Request URL issued from Cluster WebUI to the WebManager server.

IP

IP address of a client that operated Cluster WebUI.

UserName
Name of a user who executed the operation.
When a user logged in to Cluster WebUI by using the OS authentication method, the user name is output.

HTTP-Status
HTTP status code.
200: Success
Other than 200: Failure

ErrorCode

Return value of the executed operation.

ResponseTime(ms)
Time taken for executing the operation (in milliseconds).
This is outputted in milliseconds.

ServerName
Name of a server to be operated.
Its server name or IP address is outputted.
It is outputted when the name of a server to be operated is specified.

GroupName
Name of a group to be operated.
It is outputted when the name of a group to be operated is specified.

ResourceName
Name of a resource to be operated.
Outputted is the heartbeat resource name, network partition resolution resource name, group resource name, or monitor resource name.
It is outputted when the name of a resource to be operated is specified.

ResourceType
Type of a resource to be operated.
It is output when the type of a resource to be operated is specified.

Parameters...

Operation-specific parameters.

The following output is an example of the operation log file of Cluster WebUI:

"Date","Operation","Request","IP","UserName","HTTP-Status","ErrorCode","ResponseTime(ms)","ServerName","GroupName","ResourceName","ResourceType","Parameters..."
"2020/08/14 17:08:39.902","Cluster properties","/GetClusterproInfo.js","10.0.0.15","user1",200,0,141,,,,
"2020/08/14 17:08:46.659","Monitor properties","/GetMonitorResourceProperty.js","10.0.0.15","user1",200,0,47,,,"fipw1","fipw"
"2020/08/14 17:15:31.093","Resource properties","/GetGroupResourceProperty.js","10.0.0.15","user1",200,0,47,,"failoverA","fip1","fip"
"2020/08/14 17:15:45.309","Start group","/GroupStart.js","10.0.0.15","user1",200,0,0,"server1","failoverA",,
"2020/08/14 17:16:23.862","Suspend all monitors","/AllMonitorSuspend.js","10.0.0.15","user1",200,0,453,"server1",,,,"server2"
                                                    :

The following is an example of the operation log file of Cluster WebUI outputted when the authentication fails:

  • When the cluster password method is used

    "Date","Operation","Request","IP","UserName","HTTP-Status","ErrorCode","ResponseTime(ms)","ServerName","GroupName","ResourceName","ResourceType","Parameters..."
"2020/11/20 09:29:59.710","Login","/Login.js","10.0.0.15","",403,,0,,,,

  • When the OS authentication method is used

    "Date","Operation","Request","IP","UserName","HTTP-Status","ErrorCode","ResponseTime(ms)","ServerName","GroupName","ResourceName","ResourceType","Parameters..."
"2020/11/20 09:29:59.710","Login User","/LoginUser.js","10.0.0.15","user1",401,,0,,,,

2.8. Function for outputting an API service operation log file

With the Output API Service Operation Log checkbox checked in the API tab of Cluster Properties in the config mode of Cluster WebUI, a log file is outputted containing information handled by the RESTful API. This CSV-format file is hereinafter called an API service operation log file.

restapiope.cur
restapiope.pre<x>

cur

Indicates the last outputted log file.

pre<x>
Indicates a previously outputted but rotated log file.
pre, pre1, pre2, ..., in reverse chronological order.
When the prescribed number of existing log files is exceeded, the oldest log file is deleted.

Where to save

Directory as Log output path in the config mode of Cluster WebUI

The operation information to be outputted includes the following items:

Item name

Description

Date
Time when the operation information is outputted.
This is outputted in the form below (000 in milliseconds):
YYYY/MM/DD HH:MM:SS.000

Method

Either of the following HTTP request methods: GET or POST.

Request

Issued request-URI.

IP

IP address of the client which issued the request.

UserName

Name of a user who executed the operation.

HTTP-Status
HTTP status code.
200: Success
Other than 200: Failure

ErrorCode

Return value of the executed operation.

ResponseTime(ms)
Time taken for executing the operation (in milliseconds).
This is outputted in milliseconds.

Here is an example of the contents of an outputted API service operation log file:

"Date","Method","Request","IP","UserName","HTTP-Status","ErrorCode","ResponseTime(ms)"
"2023/05/28 16:34:08.007","GET","https://10.0.0.1:29009/api/v1/cluster","10.0.0.15","user1",200,0,84
"2023/05/28 16:34:08.007","GET","https://10.0.0.1:29009/api/v1/servers/servers?select=name","10.0.0.15","user1",200,0,84
"2023/05/28 16:35:03.283","POST","https://10.0.0.1:29009/api/v1/cluster/start","10.0.0.15","user1",200,0,142
"2023/05/28 16:35:03.283","POST","https://10.0.0.1:29009/api/v1/groups/failoverA/start -d '{ "target" : "server1" }'","10.0.0.15","user1",200,0,142
"2023/05/28 16:35:03.283","POST","https://10.0.0.1:29009/api/v1/resources/fip1/start -d '{ "target" : "server1" }'","10.0.0.15","user1",200,0,142
"2023/05/28 16:35:03.283","POST","https://10.0.0.1:29009/api/v1/monitors/fipw1/suspend -d '{ "target" : "server1" }'","10.0.0.15","root",200,0,142
                                                    :

2.9. Function for obtaining a log file for investigation

If an activation failure occurred in a group/monitor resource or a forced-stop resource failed in a forced stop, such information is collected and saved as a compressed file to the following directory: <installation path>/log/ecap. The format of the file name is <date and time when the event occurred>_<module name>_<event ID>.tar.gz.

You can obtain this log file through Cluster WebUI. To do so, in the config mode of Cluster WebUI, go to Cluster Properties -> the Alert Log tab, then check the Enable a log file for investigation to be downloaded.

The compressed file contains the output of an executed command shared by resource types and that of one specific to a resource type.

  • Output of an executed command shared by resource types

  • Output of an executed command specific to a resource type

    • The output is stored as a text file in Markdown format: <resource type>.ecap.md.

    • This is outputted by executing the following command specific to a resource type (even if this command does not exist, the command shared by resource types is run):

      Resource type

      Command name

      Necessary package

      Floating IP resource

      ip n

      iproute

      ping -w 3 <the IP address>

      iputils

      Dynamic DNS resource

      nslookup -timeout=3 <the virtual host name>

      bind-utils

      dig any +time=3 <the virtual host name>

      bind-utils

      NIC Link Up/Down monitor resource

      ethtool <the name of the NIC interface>

      ethtool

      Floating IP monitor resource

      ip n

      iproute

      ping -w 3 <the IP address>

      iputils

      Dynamic DNS monitor resource

      nslookup -timeout=3 <the virtual host name>

      bind-utils

      dig any +time=3 <the virtual host name>

      bind-utils

Note

The log file for investigation may not be appropriately obtained, if the same event and the same module occurred more than once at the same period of time.

2.10. Communication ports

EXPRESSCLUSTER uses several port numbers. Change the firewall settings so that EXPRESSCLUSTER can use some port numbers.

For a cloud environment, allow access to ports numbered as below, not only in a firewall configuration at the instance side but also in a security configuration at the cloud infrastructure side.

Refer to "Getting Started Guide" > "Notes and Restrictions" > "Communication port number" for port numbers used for EXPRESSCLUSTER.

2.11. Cluster driver device information

  • The mirror driver mainly uses 218 as the major number. Make sure that no other driver uses this major number. However, this major number can be changed to avoid using 218 due to system restrictions.

  • The kernel mode LAN heartbeat driver uses 10 as the major number, and mainly uses 253 as the minor number. Make sure that no other driver uses these major and minor numbers.

  • The keepalive driver uses 10 as the major number, and mainly uses 254 as the minor number. Make sure that no other driver uses these major and minor numbers.

2.12. What causes servers to shut down

When any one of the following errors occurs, EXPRESSCLUSTER shuts down, resets servers, or performs panic of servers to protect resources.

2.12.1. Final action for an error in resource activation or deactivation

When the final action for errors in resource activation/deactivation is specified as one of the following:

Final action

Result

The cluster service stops and the OS shuts down.

Causes normal shutdown after the group resources stop.

The cluster service stops and the OS reboots.

Causes normal reboot after the group resources stop.

Sysrq Panic

Performs a panic upon group resource activation/deactivation error.

Keepalive Reset

Performs a reset upon group resource activation/deactivation error.

Keepalive Panic

Performs a panic upon group resource activation/deactivation error.

BMC Reset

Performs a reset upon group resource activation/deactivation error.

BMC Power Off

Performs a power off upon group resource activation/deactivation error.

BMC power Cycle

Performs a power cycle upon group resource activation/deactivation error.

BMC NMI

Causes NMI upon group resource activation/deactivation error.

2.12.2. Action for resource activation or deactivation stall generation

When one of the following is specified as the final action to be applied upon the occurrence of an error in resource activation/deactivation, and if resource activation/deactivation takes more time than expected:

Action performed when a stall occurs

Result

The cluster service stops and the OS shuts down.

When a group resource activation/deactivation stall occurs, performs normal shutdown after the group resources stop.

The cluster service stops and the OS reboots.

When a group resource activation/deactivation stall occurs, performs normal reboot after the group resources stop.

Sysrq Panic

When a group resource activation/deactivation stall occurs, performs a panic.

Keepalive Reset

When a group resource activation/deactivation stall occurs, performs a reset.

Keepalive Panic

When a group resource activation/deactivation stall occurs, performs a panic.

BMC Reset

When a group resource activation/deactivation stall occurs, performs a reset.

BMC Power Off

When a group resource activation/deactivation stall occurs, performs a power off.

BMC power Cycle

When a group resource activation/deactivation stall occurs, performs a power cycle.

BMC NMI

When a group resource activation/deactivation stall occurs, performs an NMI.

The OS shuts down if the resource activation or deactivation takes an unexpectedly long time. The OS shuts down, regardless of the setting of recovery in the event of a resource activation or deactivation error.

If a resource activation stall occurs, alert occurs and the following message is output to syslog.

  • Module type: rc

  • Event ID: 32

  • Message: Activating %1 resource has failed.(99 : command is timeout)

  • Description: Failed to activate 1 resource.

If a resource deactivation stall occurs, alert occurs and the following message is output to syslog.

  • Module type: rc

  • Event ID: 42

  • Message: Stopping %1 resource has failed.(99 : command is timeout)

  • Description: Failed to stop the %1 resource.

2.12.3. Final action at detection of an error in monitor resource

When the final action for errors in monitor resource monitoring is specified as one of the following:

Final action

Result

Stop cluster service and shut down the OS

Causes shutdown after the group resources stop.

Stop cluster service and reboot the OS

Causes reboot after the group resources stop.

Sysrq Panic

Causes panic when an error is detected in monitor resource.

Keepalive Reset

Causes reset when an error is detected in monitor resource.

Keepalive Panic

Causes panic when an error is detected in monitor resource.

BMC Reset

Causes reset when an error is detected in monitor resource.

BMC Power Off

Causes power off when an error is detected in monitor resource.

BMC Power Cycle

Causes power cycle when an error is detected in monitor resource.

BMC NMI

Causes NMI when an error is detected in monitor resource.

2.12.4. Forced stop action

When the type of forced stop is configured as BMC:

Forced stop action

Result

BMC reset

Causes reset in the failing server in which a failover group existed.

BMC power off

Causes power off in the failing server in which a failover group existed.

BMC power cycle

Causes power cycle in the failing server in which a failover group existed.

BMC NMI

Causes NMI in the failing server in which a failover group existed.

When the type of forced stop is configured as vCenter:

Forced stop action

Result

Power off

Causes power off in the failing server in which a failover group existed.

Reset

Causes reset in the failing server where the failover group existed.

When the type of forced stop is configured as AWS or OCI:

Forced stop action

Result

stop

Stops the instance of the failing server where the failover group existed.

reboot

Reboots the instance of the failing server where the failover group existed.

When the type of forced stop is configured as Azure:

Forced stop action

Result

stop and deallocate

Stops the instance of the failing server where the failover group existed. 6

stop only

Stops the instance of the failing server where the failover group existed. 7

reboot

Reboots the instance of the failing server where the failover group existed.
6

Stops the instance through the shutdown sequence. Since allocated resources (e.g., public IP address) are also released, no charging occurs.

7

Stops the instance not through the shutdown sequence. Since allocated resources are maintained, the charging continues.

2.12.5. Emergency server shutdown, emergency server reboot

When an abnormal termination is detected in any of the following processes, a shutdown or reboot is generated after the group resource stops. Shutdown or reboot generation depends on the setting of Action When the Cluster Service Process Is Abnormal.

  • clprc

  • clprm

2.12.6. Resource deactivation error in stopping the EXPRESSCLUSTER daemon

If there is a failure to deactivate the resource during the EXPRESSCLUSTER daemon stop process, the action set in [Action When the Cluster Service Process Is Abnormal] is executed.

2.12.7. Stall detection in user space

When a server stalls longer than the heartbeat time-out, an OS hardware reset, panic, or I/O fencing is generated. Hardware reset or panic generation depends on the setting of Operation at Timeout Detection of the user-mode monitor resource.

2.12.8. Stall detection during shutdown process

When a server stalls during the OS shutdown process, an OS hardware reset, panic, or I/O fencing is generated. Hardware reset or panic generation depends on the setting of Operation at Timeout Detection of the shutdown monitor.

2.12.9. Recovery from network partitioning

When any network partition resolution resources are not set, if all heartbeats are disrupted (network partitioning), both servers failover to each other. As a result, groups are activated on both servers. Even when network partition resolution resources are set, groups may be activated on both servers.

If interconnections are recovered from this condition, EXPRESSCLUSTER causes shutdown on both or one of the servers.

For details of network partitioning, see "When network partitioning occurs" in "Troubleshooting" in the "Reference Guide".

2.12.10. Network partition resolution

In a cluster system where network partition resolution resources are configured, the network partition resolution is performed when all heartbeats are interrupted (network partition). If this is determined to be caused by the network partitions, some or all of the servers are shut down or stop their services. Shutdown or service stop generation depends on the setting of Action at NP Occurrence.

For details on the network partition resolution, see "Network partition resolution resources details" in the "Reference Guide".

2.12.11. Mirror disk error ~For Replicator~

When an error occurs in a mirror disk, the mirror agent causes reset.

2.12.12. Hybrid disk error ~For Replicator DR~

When an error occurs in a hybrid disk, the mirror agent causes reset.

2.12.13. Failure in suspending or resuming the cluster

If suspending or resuming the cluster fails, the server is shut down.

2.13. Configuring the settings to temporarily prevent execution of failover

Follow the steps below to temporarily prevent failover caused by a failed server from occurring.

  • Temporarily adjust time-out
    By temporarily adjusting time-out, you can prevent a failover caused by a failed server from occurring.
    The clptoratio command is used to temporarily adjust time-out. Run the clptoratio command on one of the servers in the cluster.

    (Example) To extend the heartbeat time-out to an hour, or 3600 seconds, when the heartbeat time-out is set to 90 seconds:

    clptoratio -r 40 -t 1h
    For more information on the clptoratio command, see "Adjusting time-out temporarily (clptoratio command)" in "EXPRESSCLUSTER command reference" in the "Reference Guide".
  • Releasing temporary time-out adjustment
    Releases the temporary adjustment of time-out. Execute the clptoratio command for any server in the cluster.
    clptoratio -i
    For more information on the clptoratio command, see "Adjusting time-out temporarily (clptoratio command)" in "EXPRESSCLUSTER command reference" in the "Reference Guide".

Follow the steps below to temporarily prevent failover caused by a monitor error by temporarily stopping monitor resource monitoring.

  • Suspending monitoring operation of monitor resources
    By suspending monitoring operations, a failover caused by monitoring can be prevented.
    The clpmonctrl command is used to suspend monitoring. Run the clpmonctrl command on all servers in the cluster.Another way is to use the -h option on a server in the cluster and run the clpmonctrl command for all the servers.

    (Example) To suspend all monitoring operations:on the server in which the command is run:

    clpmonctrl -s

    (Example) To suspend all monitoring operations on the server with -h option specified

    clpmonctrl -s -h <server name>
    For more information on the clpmonctrl command, see "Controlling monitor resources (clpmonctrl command)" in "EXPRESSCLUSTER command reference" in the "Reference Guide".
  • Restarting monitoring operation of monitor resources
    Resumes monitoring. Execute the clpmonctrl command for all servers in the cluster.Another way is to use the -h option on a server in the cluster and run the clpmonctrl command for all the servers.

    (Example) Resuming all monitoring operations:on the server in which the command is run:

    clpmonctrl -r

    (Example) To resume all monitoring operations on the server with -h option specified

    clpmonctrl -r -h <server name>

    For more information on the clpmonctrl command, see "Controlling monitor resources (clpmonctrl command)" in "EXPRESSCLUSTER command reference" in the "Reference Guide".

Follow the steps below to temporarily prevent failover caused by a monitor error by disabling recovery action for a monitor resource error.

  • Disabling recovery action for a monitor resource error
    When you disable recovery action for a monitor resource error, recovery action is not performed even if a monitor resource detects an error. To set this feature, check the Recovery action when a monitor resource error is detected checkbox in Disable cluster operation under the Extension tab of Cluster properties in config mode of Cluster WebUI and update the setting.
  • Not disabling recovery action for a monitor resource error
    Enable recovery action for a monitor resource error. Uncheck the Recovery action when a monitor resource error is detected checkbox in Disable cluster operation under the Extension tab of Cluster properties in config mode of Cluster WebUI and update the setting.

Follow the steps below to temporarily prevent failover caused by an activation error by disabling recovery action for a group resource activation error.

  • Disabling recovery action for a group resource activation error
    When you disable recovery action for a group resource activation error, recovery action is not performed even if a group resource detects an activation error. To set this feature, check the Recovery operation when a group resource activation error is detected checkbox in Disable cluster operation under the Extension tab of Cluster properties in config mode of Cluster WebUI and update the setting.
  • Not disabling recovery action for a group resource activation error
    Enable recovery action for a group resource activation error. Uncheck the Recovery operation when a group resource activation error is detected checkbox in Disable cluster operation under the Extension tab of Cluster properties in config mode of Cluster WebUI and update the setting.

2.14. How to replace a mirror disk with a new one

When the replacement of mirror disks is necessary due to mirror disk breakdown or some reasons after starting operation, run the following steps:

See also

For details on how to stop and start daemons, see "Suspending EXPRESSCLUSTER" in "Preparing to operate a cluster system" in the "Installation and Configuration Guide".

2.14.1. In case of replacing a mirror disk constructed with a single disk(non-RAID)

  1. Stop the server of which the mirror disk is going to be replaced.

    Note

    Before shutting down the server, it is recommended that the steps in "Disabling the EXPRESSCLUSTER daemon" in the "Installation and Configuration Guide" be executed.
    On the target server, execute the following command to disable the daemon.
    clpsvcctrl.sh --disable core mgr

    • If a hybrid disk failure occurs, terminate all servers connected to the disk to be replaced.

  2. Install a new disk in the server.

  3. Start up the server in which the new disk was installed. At this time, change the setting so that the EXPRESSCLUSTER services will not be executed. In case of not having disabled the EXPRESSCLUSTER daemon in the step 1, the daemons start up on run level 1 at OS startup.

  4. Construct the same partition as the original disk to the new disk by fdisk command.

    Note

  5. Prevent initial mirror construction from being performed automatically.

    • (A) In the state in which the operation is being performed on the server on which a mirror disk is not replaced (state in which the group containing mirror disk resources is active), you want to concurrently perform disk copy (initial mirror construction), there is no particular need to make sure that initial mirror construction is not automatically performed.

    • (B) If the operation could be stopped until disk copy is completed (the group may be deactivated), deactivate the group containing the mirror disk resource.

    Note

    • With procedure (A), copy is performed by the amount equal to that of disk space used, depending on the type of file system, so the copy time may depend on the amount of disk space used.
      Also, because the operation and copy are performed concurrently, the load may become high and copy may take time depending on the case.

    • With procedure (B) whereby disk copy is performed while the operation is stopped (the group is deactivated), copy is performed by the amount equal to that of disk space used, depending on the file system, so the copy time may depend on the amount of disk space used. The operation (group activation) can be started after the completion of copy.

  6. On the server on which a new disk has been installed, enable the EXPRESSCLUSTER daemon, and restart the server.

    Note

    • In case that the steps in "Disabling the EXPRESSCLUSTER daemon" in the Installation and Configuration Guide were executed before shutting down the server, enable the EXPRESSCLUSTER daemons at this time.
      On the target server, execute the following command to enable the daemon.
      clpsvcctrl.sh --enable core mgr

  7. Start the initial mirror construction (disk copy) by executing the command described below.

    • (A) When performing an operation on a server on which the mirror disk has not been replaced
      The initial mirror construction (disk copy) is automatically started.
      If you set Execute the initial mirror construction to Off, construction is not started automatically; use Mirror Disks or either of the following commands to start it manually

      [For a mirror disk]

      clpmdctrl --force copy_source_server_name> <mirror_disk_resource_name>

      [For a hybrid disk]

      clphdctrl --force copy_source_server_name> <hybrid_disk_resource_name>
    • (B) If the operation is stopped, and the operation is to be started after the completion of disk copy
      (When performing copy when the group containing the mirror disk resource is deactivated)

      [For a mirror disk]

      clpmdctrl --force <copy_source_server_name> <mirror_disk_resource_name>

      [For a hybrid disk]

      clphdctrl --force <copy_source_server_name> <hybrid_disk_resource_name>
  8. If initial mirror construction is started while the operation is stopped (deactivated) (B), you can start the operation (activate the group) after the completion of the initial mirror construction (after the completion of disk copy).
    If mirror recovery is interrupted, start initial mirror construction without activating the group.

2.14.2. In case of replacing a mirror disk constructed with a number of disks(RAID)

  1. Stop the server of which the mirror disks are going to be replaced.

    Note

    • Before shutting down the server, it is recommended that the steps in "Disabling the EXPRESSCLUSTER daemon" in the Installation and Configuration Guide be executed.
      On the target server, execute the following command to disable the daemon.
      clpsvcctrl.sh --disable core mgr

    • If a hybrid disk failure occurs, terminate all servers connected to the disk to be replaced.

  2. Install the new disks in the server.

  3. Start up the server.

  4. Reconstruct the RAID before OS startup.

  5. Change the setting so that the EXPRESSCLUSTER services will not be executed at OS startup. In case of not having disabled the EXPRESSCLUSTER daemon in the step 1, startup the daemons on run level 1 at OS startup, then startup the daemons on run level 3 after disabling the daemons.
    Back up data from the data partition as required.

  6. If LUN is initialized, use the fdisk command to create cluster and data partitions on a new disk.

    Note

    • If a hybrid disk failure occurs, terminate all servers connected to the disk to be replaced.

  7. Login as the root and initialize the cluster partition using one of the following methods.

    • Method (1) Without using the dd command

      For the mirror disk

      clpmdinit --create force <mirror disk resource name>

      For the hybrid disk

      clphdinit --create force <hybrid disk resource name>

      Note

      • For the mirror disk, if Execute initial mkfs is set to "on" when the mirror disk resource is set up, mkfs is executed upon execution of this command to initialize the file system.
        However, mkfs may take a long time to complete in the case of a large-capacity disk. (once mkfs is executed, any data saved in the data partition will be erased. Back up the data in the data partition as required, therefore, before executing this command.)
        Mirror data is copied from the destination server by means of the entire recovery described later.

      • If a hybrid disk failure occurs, terminate all servers connected to the disk to be replaced.

    • Method (2) Using the dd command

      For the mirror disk

      dd if=/dev/zero of=<cluster partition device name (Example: /dev/sdb1)>
clpmdinit --create quick <mirror disk resource name>

      For the hybrid disk

      dd if=/dev/zero of=<cluster partition device name (Example: /dev/sdb1)>
clphdinit --create quick <hybrid disk resource name>

    Note

    • When the dd command is executed, data in the partition specified by of= is initialized. Confirm whether the partition device name is correct, and then execute the dd command.

    • When the dd command is executed, the following message may appear. This does not, however, indicate an error.
      dd: writing to <CLUSTER partition device name>: No space left on device

    • Mirror data is copied from the destination server by means of the entire recovery described later. Back up the data in the data partition as required, therefore, before executing this command.

    • If a hybrid disk failure occurs, terminate all servers connected to the disk to be replaced.

  8. Prevent initial mirror construction from being performed automatically.

    • (A) In the state in which the operation is being performed on the server on which a mirror disk is not replaced (state in which the group containing mirror disk resources is active), you want to concurrently perform disk copy (initial mirror construction), there is no particular need to make sure that initial mirror construction is not automatically performed.

    • (B) If the operation could be stopped until disk copy is completed (the group may be deactivated), deactivate the group containing the mirror disk resource.

    Note

    • With procedure (A), copy is performed by the amount equal to that of disk space used, depending on the type of file system, so the copy time may depend on the amount of disk space used.
      Also, because the operation and copy are performed concurrently, the load may become high and copy may take time depending on the case.

    • With procedure (B) whereby disk copy is performed while the operation is stopped (the group is deactivated), copy is performed by the amount equal to that of disk space used, depending on the file system, so the copy time may depend on the amount of disk space used. The start of the operation (group activation) can be performed after the completion of copy.

  9. On a server on which a disk has been replaced, enable the EXPRESSCLUSTER daemon, and then restart the server.

    Note

    • In the case that the steps in "Disabling the EXPRESSCLUSTER daemon" in the "Installation and Configuration Guide" were executed before shutting down the server, enable the EXPRESSCLUSTER daemons at this time.
      On the target server, execute the following command to enable the daemon.
      clpsvcctrl.sh --enable core mgr

  10. Use the following command to start the initial mirror construction (disk copy).

    • (A) When performing an operation on a server on which the mirror disk has not been replaced

      The initial mirror construction (disk copy) is automatically started.
      If you set Execute the initial mirror construction to Off, construction is not started automatically; use Mirror Disks or either of the following commands to start it manually

      [For a mirror disk]

      clpmdctrl --force <copy_source_server_name> <mirror_disk_resource_name>

      [For a hybrid disk]

      clphdctrl --force <copy_source_server_name> <hybrid_disk_resource_name>
    • (B) If the operation is stopped, and is to be started after disk copy has been completed
      (When performing copy in the state in which the group containing the mirror disk resource is deactivated)

      [For a mirror disk]

      clpmdctrl --force <copy_source_server_name> <mirror_disk_resource_name>

      [For a hybrid disk]

      clphdctrl --force <copy_source_server_name> <hybrid_disk_resource_name>
  11. If initial mirror construction is started while the operation is stopped (deactivated) (B), you can start the operation (activate the group) after the completion of the initial mirror construction (after the completion of disk copy).
    If mirror recovery is interrupted, start the initial mirror construction without activating the group.

2.14.3. In case of replacing mirror disks of both servers

Note

The data of mirror disks are lost after replacing the mirror disks of both servers. Restore the data from backup data or other media as necessary after replacing the disks.

  1. Stop the both servers.

    Note

    • Before shutting down both servers, it is recommended that the steps in "Disabling the EXPRESSCLUSTER daemon" in the Installation and Configuration Guide are executed.
      On the target server, execute the following command to disable the daemon.
      clpsvcctrl.sh --disable core mgr

  2. Install the new disks in both servers.

  3. Startup both servers. At this time, change the setting so that the EXPRESSCLUSTER services will not be executed. In case of not having disabled the EXPRESSCLUSTER daemon in the step 1, the daemons start up on run level 1 at OS startup.

  4. Construct the same partition as the original disk to the new disks of both servers by fdisk command.

    Note

  5. Restart both servers.

    Note

    • In the case that the steps in "Disabling the EXPRESSCLUSTER daemon" in the "Installation and Configuration Guide" were executed before shutting down the server, enable the EXPRESSCLUSTER daemons at this time.
      On the target server, execute the following command to enable the daemon.
      clpsvcctrl.sh --enable core mgr
  6. The initial mirror construction (entire mirror recovery) starts automatically by restarting.
    If you set Execute the initial mirror construction to Off, the normal state is assumed directly without automatically starting. Thus, in this case, use the Mirror Disks of Cluster WebUI, clpmdctrl, or the clphdctrl command to manually start full mirror recovery.

  7. After the completion of full mirror recovery, recover the data from a backup or the like after the completion of full mirror recovery.

2.15. How to replace a server with a new one ~For a shared disk~

Connect to the Cluster WebUI with a management IP address. If you do not have any management IP address, connect to it by using the IP address of a server that is not to be replaced.

  1. Install the EXPRESSCLUSTER Server to the new server.

For details, see "Installing EXPRESSCLUSTER" in the Installation and Configuration Guide. The server on which you installed the EXPRESSCLUSTER Server should be restarted after the installation.

  1. Upload the cluster configuration data in config mode of Cluster WebUI you connected to.

    If you use a fixed term license, run the following command:

    clplcnsc --reregister <a folder path for saved license files>

  2. Restart the replaced server.

2.16. How to replace a server with a new one ~For a mirror disk~

2.16.1. Replacing a server and its mirror disk

Connect to the Cluster WebUI with a management IP address. If you do not have any management IP address, connect to it by using the IP address of a server that is not to be replaced.

  1. Replace the failed server machine and the disk. Set the same IP address and host name in the new server as the old server.

    Normal Server 1 with a disk connected and unbootable Server 2 with a disk connected

    Fig. 2.2 Not able to start Server 2 or use its disk

    Normal Server 1 with a disk connected and new Server 2 with a new disk connected

    Fig. 2.3 Replacing Server 2 and its disk with a new server and a new disk

  2. Create partitions in the new disk by executing the fdisk command.
    Install the EXPRESSCLUSTER Server on the new server. For details, see "Installing EXPRESSCLUSTER" in the Installation and Configuration Guide. The server on which you installed the EXPRESSCLUSTER Server should be restarted after the installation.
    Server 1, and Server 2 on which the fdisk command is executed

    Fig. 2.4 Creating partitions in the new disk

  3. When using the disk that was used as a mirror disk before, initialize the cluster partition.

    A disk with the initialized cluster partition, and with the data partition on which a file system is created

    Fig. 2.5 Partitioning the new disk

  4. Upload the cluster configuration data in the config mode of Cluster WebUI you connected to. When uploading the data completes, restart the replaced server.

    If you use a fixed term license, run the following command:

    clplcnsc --reregister <a folder path for saved license files>
  5. After the server is restarted, the cluster partitions in the new disk will be initialized and a file system will be created in the data partition.
    The mirror recovery is executed if the initial mirror construction is set. If not, you have to manually recover mirroring.
    For information on recovery of disk mirroring, refer to "Recovering mirror with a command" and "Recovering mirror using the Cluster WebUI" of "Troubleshooting" in "Reference Guide"
    In mirror recovery, the data is fully copied.
    Confirm that mirroring is successfully recovered by using the WebManager or by running the following command. For details, see "Mirror-related commands" in "EXPRESSCLUSTER command reference" in the "Reference Guide".
    clpmdstat --mirror <mirror_disk_resource_name (Example: md1)>
    Data being copied from the disk of Server 1 to the disk of Server 2

    Fig. 2.6 Starting mirror recovery on Server 1 (full copy)

2.16.2. Using the mirror disk of the failed server

Connect to the Cluster WebUI with a management IP address. If you do not have any management IP address, connect to it by using the IP address of a server that is not to be replaced.

  1. Replace the failed server machine but continue using the mirror disk of the failed server. Set the same IP address and host name in the new server as before.

    Normal Server 1 with a disk connected and unbootable Server 2 with a disk connected

    Fig. 2.7 Not able to start Server 2

    Normal Server 1 with a disk connected and new Server 2

    Fig. 2.8 Replacing Server 2 with a new one

    Install the EXPRESSCLUSTER Server on the new server. For details, see "Installing EXPRESSCLUSTER" in the "Installation and Configuration Guide". Restart the server on which the EXPRESSCLUSTER Server was installed.

  2. Upload the cluster configuration data in the config mode of Cluster WebUI you connected to. When uploading the data completes, restart the replaced server.

    If you use a fixed term license, run the following command:

    clplcnsc --reregister <a folder path for saved license files>
  3. If there is no difference in mirror disks, you can immediately start the operation after restarting the server. On the other hand, if there is any difference in mirror disks, you have to recover the mirroring data after restarting the server.
    The disk mirroring is automatically recovered when auto-mirror recovery is enabled. If not, you have to manually recover disk mirroring. For information on recovery of disk mirroring, refer to "Recovering mirror with a command" and "Recovering mirror using the Cluster WebUI" of "Troubleshooting" in "Reference Guide".
    Confirm that mirroring is successfully recovered by using the Cluster WebUI or by running the following command. For details, see "Mirror-related commands" in "EXPRESSCLUSTER command reference" in the "Reference Guide".
    clpmdstat --mirror <mirror_disk_resource_name (Example: md1)>
    Data being copied from the disk of Server 1 to the disk of Server 2

    Fig. 2.9 Starting mirror recovery on Server 1 (differential copy)

2.17. How to replace a server with a new one ~For a hybrid disk~

2.17.1. Replacing a server and its non-shared hybrid disk

Connect to the Cluster WebUI with a management IP address. If you do not have any management IP address, connect to it by using the IP address of a server that is not to be replaced.

  1. Replace the failed server machine and the disk. Set the same IP address and host name in the new server as the old server.

    Normal Server 1 and Server 2 both with a shared disk connected, and unbootable Server 3 with another disk connected

    Fig. 2.10 Not able to start Server 3 or use its disk

    Normal Server 1 and Server 2 both with a disk connected, and new Server 3 with a new disk connected

    Fig. 2.11 Replacing Server 3 and its disk with a new server and a new disk

  2. Create partitions in the new disk by executing the fdisk command.

    Server 1, Server 2, and Server 3 on which the fdisk command is executed

    Fig. 2.12 Creating partitions in a new disk of Server 3

  3. Install the EXPRESSCLUSTER Server on the new server. For details, see "Installing EXPRESSCLUSTER" in the Installation and Configuration Guide. The server on which you installed the EXPRESSCLUSTER Server should be restarted after the installation.

  4. Upload the cluster configuration data in the config mode of Cluster WebUI you connected to.

    If you use a fixed term license, run the following command:

    clplcnsc --reregister <a folder path for saved license files>

  5. Execute the clphdinit command in the replaced server.

    # clphdinit --create force <Hybrid disk resource name (Example: hd1)>

  6. Restart the replaced server.

  7. After the server is restarted, the mirror recovery is executed if the initial mirror construction is set. If not, you have to manually recover mirroring.
    For information on recovery of disk mirroring, refer to "Recovering mirror with a command" and "Recovering mirror using the Cluster WebUI" of "Troubleshooting" in "Reference Guide".
    In mirror recovery, the data is fully copied.
    Confirm that mirroring is successfully recovered by using the Cluster WebUI or by running the following command. For details, see "Hybrid-disk-related commands" in "EXPRESSCLUSTER command reference" in the "Reference Guide".
    clphdstat --mirror <hybrid_disk_resource_name (Example: hd1)>
    Data being copied from the shared disk connected to Server 1 to the disk of Server 3

    Fig. 2.13 Starting mirror recovery on Server 1 (full copy)

2.17.2. Replacing a server and a hybrid disk of the shared disk

Connect to the Cluster WebUI with a management IP address. If you do not have any management IP address, connect to it by using the IP address of a server that is not to be replaced.

  1. Set the EXPRESSCLUSTER service not to start on the failed server and the other server connecting to the same shared disk.

    clpsvcctrl.sh --disable core
  2. Shut down the server that was connected to the failing server via the shared disk by running the OS shutdown command etc.
    If you want to keep the operation during replacement, move the group to server 3.
    Unbootable Server 1 with a shared disk connected, Server 2 with the same shared disk connected, and normal Server 3 with a disk connected

    Fig. 2.14 Not able to start Server 1 or use the shared disk

    Unbootable Server 1 with a shared disk connected, Server 2 that is shut down and with the same shared disk connected, and normal Server 3 with a disk connected

    Fig. 2.15 Shutting down Server 2

  3. Replace the failed server machine and the shared disk. Set the same IP address and host name in the new server as the old server.

    Replaced Server 1 with a replaced shared disk connected, Server 2 with the same shared disk connected, and normal Server 3 with a disk connected

    Fig. 2.16 Replacing Server 1 and the shared disk with a new server and a new disk

  4. Create disk partitions from the replaced server by executing the fdisk command.

    Server 1 on which the fdisk command is executed, Server 2 that is shut down, and normal Server 3

    Fig. 2.17 Creating partitions in a new shared disk connected to Server 1

  5. Install the EXPRESSCLUSTER Server on the new server. For details, see "Installing EXPRESSCLUSTER" in the "Installation and Configuration Guide". The server on which you installed the EXPRESSCLUSTER Server should be restarted after the installation. Start the server that was connected to the failing server via the shared disk.

    EXPRESSCLUSTER is not started on the non-replaced server of the servers connected to the shared disk.
    Restarted Server 1 and started Server 2 after installing EXPRESSCLUSTER, and Server 3 in operation

    Fig. 2.18 Installing EXPRESSCLUSTER on Server 1 and starting Server 2

  6. Upload the cluster configuration data in the config mode of Cluster WebUI you connected to.

    If you use a fixed term license, run the following command:

    clplcnsc --reregister <a folder path for saved license files>
  7. On the replaced server, run the clphdinit command.
    # clphdinit --create force <hybrid disk resource name(example: hd1)>

  8. Set the EXPRESSCLUSTER service to start on the failed server and the other server connecting to the same shared disk.

    clpsvcctrl.sh --enable core

  9. Restart the replaced server as well as the server that was connected to the failing server via the shared disk.

    Server 1 and Server 2 both restarted after changing the settings, and normal Server 3

    Fig. 2.19 Restarting the replaced server as well as the server that was connected to the failing server via the shared disk

  10. After the server is restarted, the mirror recovery is executed if the initial mirror construction is set. If not, you have to manually recover mirroring
    For information on recovery of disk mirroring, refer to "Recovering mirror with a command" and "Recovering mirror using the Cluster WebUI" of "Troubleshooting" in "Reference Guide".
    The destination server of disk mirroring is the current server of the server group to which the shared disk is connected (The figure below shows an example where the server 1 is the current server).
    In mirror recovery, the data is fully copied.
    Check that mirror recovery has completed by running the following command, or by using WebManager. For details, see "Hybrid-disk-related commands" in "EXPRESSCLUSTER command reference" in the "Reference Guide".
    clphdstat --mirror <hybrid disk resource name (example: hd1)>
    Data being copied from the shared disk connected to Server 1 to the disk of Server 3

    Fig. 2.20 Starting mirror recovery (full copy)

2.17.3. Using the disk of the failed server

Connect to the Cluster WebUI with a management IP address. If you do not have any management IP address, connect to it by using the IP address of a server that is not to be replaced.

  1. Replace the failed server machine but continue using the disk of the failed server. Set the same IP address and host name in the new server as before.

    Normal Server 1 and Server 2 both with the same shared disk connected, and unbootable Server 3 with another disk connected

    Fig. 2.21 Not able to start Server 1 or use the shared disk

    Normal Server 1 and Server 2 both with the same shared disk connected, and replaced Server 3 with another disk connected

    Fig. 2.22 Replacing Server 3 with a new one

  2. Install the EXPRESSCLUSTER Server on the new server. For details, see "Installing EXPRESSCLUSTER" in the Installation and Configuration Guide. Restart the server on which the EXPRESSCLUSTER Server was installed.

  3. Upload the cluster configuration data in the config mode of Cluster WebUI you connected to. When uploading the data completes, restart the replaced server.

    If you use a fixed term license, run the following command:

    clplcnsc --reregister <a folder path for saved license files>
  4. If there is no difference in mirror disks, you can immediately start the operation after restarting the server. On the other hand, if there is any difference in mirror disks, you have to recover the mirroring data after restarting the server.
    The disk mirroring is automatically recovered when auto-mirror recovery is enabled. If not, you have to manually recover disk mirroring. For information on recovery of disk mirroring, refer to "Recovering mirror with a command" and "Recovering mirror using the Cluster WebUI" of "Troubleshooting" in "Reference Guide".
    Confirm that mirroring is successfully recovered by using the Cluster WebUI or by running the following command. For details, see "Hybrid-disk-related commands" in "EXPRESSCLUSTER command reference" in the "Reference Guide".
    clpmdstat --mirror <hybrid_disk_resource_name (Example: hd1)>
    Data being copied from the shared disk connected to Server 1 to the disk of Server 3

    Fig. 2.23 Starting mirror recovery on Server 1 (differential copy)

2.17.4. Replacing a server to which the shared disk is connected

Connect to the Cluster WebUI with a management IP address. If you do not have any management IP address, connect to it by using the IP address of a server that is not to be replaced.

  1. Replace the failed server machine and the shared disk. Set the same IP address and host name in the new server as the old server.

    Unbootable Server 1 and normal Server 2 both with the same shared disk connected, and normal Server 3 with a disk connected

    Fig. 2.24 Not able to start Server 1

    Replaced Server 1 and normal Server 2 both with the same shared disk connected, and normal Server 3 with a disk connected

    Fig. 2.25 Replacing Server 1 with a new one

  2. Install the EXPRESSCLUSTER Server on the new server. For details, see "Installing EXPRESSCLUSTER" in the "Installation and Configuration Guide". Restart the server on which the EXPRESSCLUSTER Server was installed

  3. Upload the cluster configuration data in the config mode of Cluster WebUI you connected to.

    If you use a fixed term license, run the following command:

    clplcnsc --reregister <a folder path for saved license files>

    When uploading the data completes, restart the replaced server.

2.18. How to restore a virtual machine ~For a mirror disk~

If a failure occurs in the system disk of a server in a virtual environment, follow the steps below to replace the disk and to restore the contents from a backup.

Note

  • This procedure is not intended for backup/restoration by the file; but for backing up as or restoring from a disk image, outside the OS.
  • This procedure requires backing up the disk as a disk image beforehand.
    For information on how to create the disk image, refer to "2.19.1. Simultaneously backing up both active and standby mirror disks" and "2.19.3. Backing up standby mirror disks". These sections instruct you to execute clpbackup.sh --pre --no-shutdown as a step for mirror disk resources. Instead, however, execute clpbackup.sh --pre to shut down the server, and then create the backup. This is because, when you back up the system disk, it is recommended to make the system disk static.
  • This procedure is for restoring the system disk and mirror disk resources on the server, but not for separately restoring each of the resources.

  1. Move a group which has started up on the server with the system disk to be restored (hereafter referred to as the target server), if any. After moving the group, check that each group resource is normally started up.

  2. In order to prevent the automatic mirror recovery, pause all the mirror disk monitor resources on servers other than the target server, by using Cluster WebUI or executing the following clpmonctrl command:

    clpmonctrl -s -h <server name> -m <monitor resource name>

  3. Shut down the target server by executing the following the clprestore.sh command:

    clprestore.sh --pre

  4. Use the backup image of the target server to create a new virtual hard disk.

    • If the target server currently has separate virtual hard disks (one for the system disk and the other[s] for the mirror disk resource[s]), use their backup images to create their respective new virtual hard disks.

  5. Replace the existing virtual hard disk of the target server, with the new one.
    For more information on the replacement procedure, refer to the manuals or guides of virtual platforms and cloud environments.

  6. Start up the target server.

    Note

    Starting up the target server does not automatically start up the cluster service. Since you executed clpbackup.sh --pre in creating the backup, automatic startup of the cluster service is disabled.

  7. On the target server, check that the device file name of the disk after the restoration is the same as that before the restoration. If the name is different, set it as before.

  8. Execute the following clprestore.sh command to reboot the target server.

    clprestore.sh --post

  9. Open Cluster WebUI -> Mirror disks, then make a mirror recovery (full copy) of all the mirror disk resources.

    Note

    The copy source must be a server on which data to be updated exists.
    Make a full copy instead of a differential copy, because the data difference may have become invalid during the restoration process.

  10. Resume the mirror disk monitor resources on the servers other than the target server, by using Cluster WebUI or executing the following clpmonctrl command:

    clpmonctrl -r -h <server_name> -m <monitor_resource_name>

  11. Confirm that the mirror is synchronized normally, by using Cluster WebUI or by running the clpmdstat command:

    clpmdstat --mirror <md_resource_name>

    Note

    If the mirror status is GREEN for both servers, the mirror is synchronized normally.

2.19. How to back up a mirror/hybrid disk to its disk image

Perform either of the following procedures when backing up the partition (cluster partition and data partition) for a mirror/hybrid disk, to its disk image:

2.19.1.

Simultaneously backing up both active and standby mirror disks

2.19.2.

Backing up active/standby mirror disks in each server

2.19.3.

Backing up standby mirror disks

2.19.4.

Backing up mirror disks on the single server

Note

  • These procedures are not intended for per-file backup/restoration, but for disk image backup/restoration.
    These procedures are different from that for backing up files from activated mirror disks/hybrid disks or backing up files from standby mirror disks/hybrid disks by canceling the access restriction.
    For information on the per-file backup procedure, see "Installation and Configuration Guide" -> "Verifying operation" -> "Backup procedures".
  • In these procedures, backup/restoration applies to all the mirror disks and hybrid disks on the target server. These procedures are not applicable to separate backup/restoration for each resource.
  • Back up/Restore both of the cluster partition and the data partition.

    * A mirror/hybrid disk consists of a data partition to be the mirroring target, and a cluster partition to record the management information.
    For information on the cluster partition and the data partition, see "Reference Guide" -> "Group resource details" -> "Understanding Mirror disk resources" or "Understanding Hybrid disk resources".
  • If hybrid disk resources exist, it should be determined on which server the backup is performed, in each of the server groups.
  • Each of the procedures with hybrid disk resources is written as follows: Execute clpbackup.sh --pre or clpbackup.sh --post on a server of a server group first, then perform clpbackup.sh --pre --only-shutdown or clpbackup.sh --post --only-reboot on all the other servers of the server group.
    Each of the written procedures includes the current server of the server group, as a signpost for the first server of the group on which the command is executed. However, the current server does not have to be the first server.
    If the server group has only one server, it is unnecessary to execute clpbackup.sh --pre --only-shutdown or clpbackup.sh --post --only-reboot on all the other servers of the server group.

    * In each server group, a current server is responsible for the mirror data to be transmitted/received, and to be written to its disk.
    In the active server group, the current server contains the hybrid disk resource being activated.
    In the standby server group, the current server receives the mirror data, sent from the current server of the active server group, and writes such data to its mirror disk.
  • None of the above four procedures applies to a cluster environment including a server with a version earlier than 4.3 of EXPRESSCLUSTER installed.
  • When you execute the clpbackup.sh command to shut down a server, an error may occur with such a message as "Some invalid status. Check the status of cluster.", leading to a failure in the shutdown. Then wait a while before performing the clpbackup.sh command again.
  • When you execute clpbackup.sh --post, a timeout may occur for the mirror agent being started, causing an error.
    In this case, wait a while before performing the clpbackup.sh command again.

See also

For information on the clpbackup.sh command, see "Reference Guide" -> "EXPRESSCLUSTER command reference" -> "Preparing for backup to a disk image (clpbackup.sh command)".

2.19.1. Simultaneously backing up both active and standby mirror disks

This procedure is intended for simultaneously backing up both of active mirror disks and standby mirror disks.
Perform the following procedure:

  1. Confirm that the mirror is synchronized normally, by using Cluster WebUI or by running the clpmdstat / clphdstat command:

    • For mirror disk resources:

      clpmdstat --mirror <md_resource_name>

    • For hybrid disk resources:

      clphdstat --mirror <hd_resource_name>

    Note

    If the mirror status is GREEN for both servers or both server groups, the mirror is synchronized normally.
    For hybrid disk resources, confirm which is a current server in each of the active server group and the standby server group.

  2. Stop the activated failover group (the operation) by using Cluster WebUI or by running the clpgrp command.

  3. Switch the mirror disks to backup mode by running the clpbackup.sh command.

    • For mirror disk resources:

      Execute the following command on both of the active and standby servers:

      clpbackup.sh --pre --no-shutdown

    • For hybrid disk resources:

      Execute the following command on one server in both server groups:

      clpbackup.sh --pre

    Note

    After the execution, the status of mirroring is changed to that for the backup, automatic startup of the cluster service is set to disabled.
    For mirror disk resources: After the above actions are completed, the cluster service stops.
    For hybrid disk resources: After the above actions are completed, the server shuts down.

  4. For hybrid disk resources: after shutting down the server with the clpbackup.sh command, execute the following command on all the other servers:

    clpbackup.sh --pre --only-shutdown

    Note

    When the command is executed, automatic startup of the cluster service is set to disabled and the server shuts down.

  5. Execute backup on both servers.

  6. After completing the backup, return the mirror disks from backup mode to normal mode.

    • For mirror disk resources:

      Execute the following command on both of the active and standby servers:

      clpbackup.sh --post --no-reboot

    • For hybrid disk resources:

      Start all the servers.
      Then, execute the following command on one server in both server groups:
      clpbackup.sh --post

    Note

    After the execution, the mirror status returns to normal, automatic startup of the cluster service is set to enabled.
    For mirror disk resources: After the above actions are completed, the cluster service starts up.
    For hybrid disk resources: After the above actions are completed, the server reboots. The process may take time.

  7. For hybrid disk resources: When the server starts rebooting with the clpbackup.sh command, execute the following command on all the other servers:

    clpbackup.sh --post --only-reboot

    Note

    When the command is executed, automatic startup of the cluster service is set to enabled and the server reboots.

  8. After the cluster services start up on all the active and standby servers, confirm that the mirror is synchronized normally by using Cluster WebUI or by running the clpmdstat / clphdstat command.

2.19.2. Backing up active/standby mirror disks in each server

Back up the disks in each server alternately according to the following procedure as specified in "Backing up standby mirror disks".

  1. Back up the disks on the standby server as specified in "Backing up standby mirror disks".

  2. After the completion of backup, when mirror recovery is completed to synchronize the mirror disks between the active server and the standby server, move the failover group from the active server to the standby server.

  3. Back up the disks on the previously active server as specified in "Backing up standby mirror disks".

  4. After the completion of backup, when mirror recovery is completed to synchronize the mirror disks between the active server and the standby server, move the failover group as required.

2.19.3. Backing up standby mirror disks

This procedure is intended for backing up a mirror/hybrid disk to its disk image on the standby server while the active server is activated.
Perform the following procedure:

  1. Confirm that the mirror is synchronized normally by using Cluster WebUI or by running the clpmdstat / clphdstat command:

    • For mirror disk resources:

      clpmdstat --mirror <md_resource_name>

    • For hybrid disk resources:

      clphdstat --mirror <hd_resource_name>

    Note

    If the mirror status is GREEN for both servers or both server groups, the mirror is synchronized normally.
    For hybrid disk resources, confirm which is a current server in the standby server group.

  2. In order to secure the quiescent point for data being written to the mirror area, stop the failover group (operation) including mirror disk resources and hybrid disk resources by using Cluster WebUI or by running the clpgrp command.

    Note

    Stopping the failover group prevents the backup of the data being written, or the failure to be written and backed up to a mirror area due to a cache.

  3. In order to prevent the automatic mirror recovery from working, pause all the mirror disk monitor resources/hybrid disk monitor resources on both of the active server and the standby server, by using Cluster WebUI or executing the following clpmonctrl command:

    clpmonctrl -s -h <server_name> -m <monitor_resource_name>

  4. Switch the mirror disks to backup mode by running the clpbackup.sh command.

    • For mirror disk resources:

      Execute the following command on the standby server (i.e., the server to be backed up):

      clpbackup.sh --pre --no-shutdown

    • For hybrid disk resources:

      Execute the following command on one server in the standby server group:

      clpbackup.sh --pre

    Note

    After the execution, the status of mirroring is changed to that for the backup, automatic startup of the cluster service is set to disabled.
    For mirror disk resources: After the above actions are completed, the cluster service stops.
    For hybrid disk resources: After the above actions are completed, the server shuts down.

  5. For a hybrid disk, after shutting down the server with the clpbackup.sh command, execute the following command on all the other servers of the standby server group:

    clpbackup.sh --pre --only-shutdown

    Note

    When the command is executed, automatic startup of the cluster service is set to disabled and the server shuts down.

  6. If you want to restart the operation immediately, start the failover group (operation) on the active server (i.e., the server not to be backed up) by using Cluster WebUI or by running the clpgrp command.

  7. Back up the disk to its disk images on the standby server.

  8. After the completion of the backup, return the mirror disks from backup mode to normal mode.

    • For mirror disk resources:

      Execute the following command on the standby server:

      clpbackup.sh --post --no-reboot

    • For hybrid disk resources:

      Start all the servers in the standby server group.
      Then, execute the following command on one server in the standby server group:
      clpbackup.sh --post

    Note

    After the execution, the mirror status returns to normal, automatic startup of the cluster service is set to enabled.
    For mirror disk resources: After the above actions are completed, the cluster service starts up.
    For hybrid disk resources: After the above actions are completed, the server reboots. The process may take time.

  9. For a hybrid disk, execute the following command on all the other servers of the standby server group:

    clpbackup.sh --post --only-reboot

    Note

    When the command is executed, automatic startup of the cluster service is set to enabled and the server reboots.

  10. The cluster service starts up on the standby server.
    If the mirror disk monitor resources/hybrid disk monitor resources stay paused, resume them through Cluster WebUI or by executing the following clpmonctrl command:
    clpmonctrl -r -h <server_name> -m <monitor_resource_name>

  11. The failover group (operation), if remains stopped (if not restarted immediately in the previous step), is executable on the active server.

  12. Automatic mirror recovery, if enabled, synchronizes differences in mirror disks between the active server and the standby server, generated during the backup, and then the server functions normally.
    If automatic mirror recovery is not executed and the server is not working normally, manually make a mirror recovery by clicking Difference copy icon in the Mirror disks tab of Cluster WebUI or by executing the following clpmdctrl/clphdctrl command:

    • For mirror disk resources:

      clpmdctrl --recovery <md_resource_name>

    • For hybrid disk resources:

      clphdctrl --recovery <hd_resource_name>

      Note

      For hybrid disk resources, execute this command on the current server.

2.19.4. Backing up mirror disks on the single server

For the procedure of backing up mirror disks on a single active server or its server group, while the other server or its server group is stopped and the synchronization of mirroring has not been executed, execute the backup as specified in "Simultaneously backing up both active and standby mirror disks", in which "both" or "both servers" is considered as "single" or "single server" respectively.

See also

If you want to start the failover group (operation) immediately without waiting for the startup of the other server, run the following command to cancel the cluster activation synchronization wait processing:
clpbwctrl -c
The executed command causes an error if the cluster activation synchronization wait processing is timed out or not yet started.
For information on the clpbwctrl command, see "Reference Guide" -> "EXPRESSCLUSTER command reference" -> "Controlling cluster activation synchronization wait processing (clpbwctrl command)".

2.20. How to restore the mirror/hybrid disk from the disk image

Perform either of the following procedures when restoring the partition (cluster partition and data partition) from its disk image backed up as specified in "How to back up a mirror/hybrid disk to its disk image":

2.20.1.

Simultaneously restoring the mirror disks on both of the active and standby servers from the same disk image

2.20.2.

Simultaneously restoring the mirror disks on both of the active and standby servers from their respective disk images

2.20.3.

Restoring the mirror disk on the single server from the disk image

Note

  • This section describes how to restore disk images that were backed up according to "How to back up a mirror/hybrid disk to its disk image".
    These procedures are different from that for per-file restoration of activated mirror disks/hybrid disks.
    For information on the per-file restoration, see "Installation and Configuration Guide" -> "Verifying operation" -> "Restoration procedures" -> "Restoring the data on the mirror disk or the hybrid disk".
  • In these procedures, backup/restoration applies to all the mirror disks and hybrid disks on the target server. These procedures are not applicable to separate backup/restoration for each resource.
  • Back up/Restore both of the cluster partition and the data partition.

    * A mirror/hybrid disk consists of a data partition to be the mirroring target, and a cluster partition to record the management information.
    For information on the cluster partition and the data partition, see "Reference Guide" -> "Group resource details" -> "Understanding Mirror disk resources" or "Understanding Hybrid disk resources".
  • If hybrid disk resources exist, it should be determined on which server the restoration is performed, in each of the server groups.
  • Each of the procedures with hybrid disk resources is written as follows: Execute clprestore.sh --post or clprestore.sh --post --skip-copy on a server of a server group first, then perform clprestore.sh --post --only-reboot on all the other servers of the server group.
    Each of the written procedures includes the current server of the server group, as a signpost for the first server of the group on which the command is executed. However, the current server does not have to be the first server.
    If the server group has only one server, it is unnecessary to execute clprestore.sh --post --only-reboot on all the other servers of the server group.

    * In each server group, a current server is responsible for the mirror data to be transmitted/received, and to be written to its disk.
    In the active server group, the current server contains the hybrid disk resource being activated.
    In the standby server group, the current server receives the mirror data, sent from the current server of the active server group, and writes such data to its mirror disk.
  • None of the above three procedures applies to a cluster environment including a server with a version earlier than 4.3 of EXPRESSCLUSTER installed.
  • When you execute the clprestore.sh command to shut down a server, an error may occur with such a message as "Some invalid status. Check the status of cluster.", leading to a failure in the shutdown. Then wait a while before performing the clprestore.sh command again.
  • After the restoration, if an error such as "Invalid configuration file." is displayed and the server is not restarted, check to see if the configuration data is registered, or there are any problems with the installation of EXPRESSCLUSTER or the setting of the firewall.

See also

For information on the clprestore.sh command, see "Reference Guide" -> "EXPRESSCLUSTER command reference" -> "Perform the processing after restoring from a disk image (clprestore.sh command)".

2.20.1. Simultaneously restoring the mirror disks on both of the active and standby servers from the same disk image

This procedure is intended for simultaneously restoring both of active/standby mirror disks from the same mirror disk image.
This procedure allows the mirror data of the active server and that of the standby server to be the same, thus eliminating the operation of mirror recovery (full copy) after restoration.

Important

In this procedure, Execute the initial mirror construction needs to be set to disabled in advance in the setting of mirror resources/hybrid resources.
If Execute the initial mirror construction or Execute initial mkfs is enabled, an error occurs. In this case, disable the setting by using Cluster WebUI.

  1. Stop the activated failover group by using Cluster WebUI or by running the clpgrp command.

  2. Run the following command on all the active/standby servers:
    * If the OS cannot be started and the OS or EXPRESSCLUSTER needs to be reinstalled or restored, run the following command on the server where the reinstallation or the restoration was performed:
    clprestore.sh --pre

    Note

    When the command is executed, automatic startup of the cluster service is set to disabled and the server shuts down.

  3. Restore the cluster partition and the data partition on both of the active server and standby server.
    * Restore the active server and the standby server from the same disk images.

  4. After the completion of restoring both of the active server and the standby server, start all the servers.

  5. After starting the servers, check the paths to the restored cluster partition and data partition.
    If any of the paths differs from before, start Cluster WebUI, switch to Config mode, change the path setting in Details tab of the mirror disk resource/hybrid disk resource properties, and then perform Apply the Configuration File.

    Important

    Carefully specify the path. Its incorrect setting could cause the start of mirroring to fail or the corresponding partition to be destroyed.
    Should you set a wrong path leading to a failure in the start of mirroring, begin the procedure over again from step 1.
  6. Execute the following command on each of the active server and the standby server:
    * For a hybrid disk, perform this command on one server (e.g. the current server) of the active server group and on that of the standby server group:
    clprestore.sh --post --skip-copy

    Note

    When the command is executed, all the cluster partitions are initialized, automatic startup of the cluster service is set to enabled, and the server reboots.

    Note

    If Execute the initial mirror construction is enabled in the setting of mirror disk resources/hybrid disk resources, the command fails.
    In this case, set Execute the initial mirror construction to disabled by using Cluster WebUI, click Apply the Configuration File, and then execute the command again.

    If a deactivated server exists, thereby causing Apply the Configuration File to fail with Cluster WebUI, use Export to save the configuration data to the disk.
    After extracting its configuration data on the disk accessible from the server, forcibly distribute the extracted configuration data file to the server by using the clpcfctrl command.
    clpcfctrl --push -x <path_to_the_directory_containing_the_extracted_configuration_data_file_clp.conf> --force
    * After completing the distribution, you can delete the saved compressed file and the extracted configuration data file.
    * If the distribution fails for any server due to its stoppage, remember to perform the distribution to the server later to avoid inconsistency in the configuration data.

    See also

    For information on the clpcfctrl command, see "Reference Guide" -> "EXPRESSCLUSTER command reference" -> "Changing, backing up, and checking cluster configuration data (clpcfctrl command)".

    Note

    If the mirror agent is started, the cluster partition is initialized, causing the command to fail.
    In this case, after running the clprestore.sh --pre command, start the server, and run the clprestore.sh --post --skip-copy command again.

  7. For hybrid disk resources: When the server starts rebooting with the command in step 6 above, execute the following command on all the other servers of the server group:

    clprestore.sh --post --only-reboot

    Note

    When the command is executed, automatic startup of the cluster service is set to enabled and the server reboots.

  8. After both of the active/standby servers are started, check the status of mirroring by using Cluster WebUI or by running the clpmdstat / clphdstat command.
    The status of mirroring for both the active server and the standby server is "Normal" (GREEN).

    • For mirror disk resources:

      clpmdstat --mirror <md_resource_name>

    • For hybrid disk resources:

      clphdstat --mirror <hd_resource_name>
  9. If the setting of Execute the initial mirror construction is changed, restore the original setting by using Cluster WebUI as required.
    The cluster needs to be stopped when applying the configuration.

2.20.2. Simultaneously restoring the mirror disks on both of the active and standby servers from their respective disk images

This procedure is intended for simultaneously restoring both of active/standby mirror disks from the respective mirror disk image.
Perform the following procedure:

See also

For information on the procedure of restoring both of active/standby mirror disks from the same mirror disk image, see "Simultaneously restoring the mirror disks on both of the active and standby servers from the same disk image".

  1. Stop the activated failover group by using Cluster WebUI or by running the clpgrp command.

  2. Run the following command on all the active/standby servers:
    * If the OS cannot be started and the OS or EXPRESSCLUSTER needs to be reinstalled or restored, run the following command on the server where the reinstallation or the restoration was performed:
    clprestore.sh --pre

    Note

    When the command is executed, automatic startup of the cluster service is set to disabled and the server shuts down.

  3. Restore the cluster partition and the data partition on both of the active server and standby server.

  4. After restoring both of the active server and standby server, start all the servers.

  5. After the startup, confirm that the paths of the restored cluster partition and the data partition are correct.
    If any of the paths differs from before, start Cluster WebUI, switch to Config mode, change the path setting in Details tab of the mirror disk resource/hybrid disk resource properties, and then perform Apply the Configuration File.

    Important

    Carefully specify the path. Its incorrect setting could cause the start of mirroring to fail or the corresponding partition to be destroyed.
    Should you set a wrong path leading to a failure in the start of mirroring, begin the procedure over again from step 1.

    Note

    If a deactivated server exists, thereby causing Apply the Configuration File to fail with Cluster WebUI, use Export to save the configuration data to the disk.
    After extracting its configuration data on the disk accessible from the server, forcibly distribute the extracted configuration data file to the server by using the clpcfctrl command.
    clpcfctrl --push -x <path_to_the_directory_containing_the_extracted_configuration_data_file_clp.conf> --force
    * After completing the distribution, you can delete the saved compressed file and the extracted configuration data file.
    * If the distribution fails for any server due to its stoppage, remember to perform the distribution to the server later to avoid inconsistency in the configuration data.

    See also

    For information on the clpcfctrl command, see "Reference Guide" -> "EXPRESSCLUSTER command reference" -> "Changing, backing up, and checking cluster configuration data (clpcfctrl command)".

  6. Execute the following command on each of the active server and the standby server:
    * For a hybrid disk, perform this command on one server (e.g. the current server) of the active server group and on that of the standby server group:
    clprestore.sh --post

    Note

    When the command is executed, automatic startup of the cluster service is set to enabled and the server reboots.

  7. For hybrid disk resources: When the server starts rebooting with the command in step 6 above, execute the following command on all the other servers of the server group:

    clprestore.sh --post --only-reboot

    Note

    When the command is executed, automatic startup of the cluster service is set to enabled and the server reboots.

  8. After both of the active/standby servers are started, check the status of mirroring by using Cluster WebUI or by running the clpmdstat / clphdstat command.
    The status of the mirror for both the active server and the standby server is "Abnormal" (RED).

    • For mirror disk resources:

      clpmdstat --mirror <md_resource_name>

    • For hybrid disk resources:

      clphdstat --mirror <hd_resource_name>

  9. Confirm the status of the failover group by using Cluster WebUI or by running the clpstat command.

  10. Stop the failover group that failed the startup by using Cluster WebUI or by running the clpgrp command.

  11. Change the status of the mirror side to be updated to "Normal" (GREEN) by clicking Forced mirror recovery icon in the Mirror disks tab of Cluster WebUI or by executing the clpmdctrl/clphdctrl command with the --force option on the server whose status is to be "Normal" (GREEN).

    • For mirror disk resources:

      clpmdctrl --force <md_resource_name>

    • For hybrid disk resources:

      clphdctrl --force <hd_resource_name>

  12. On the latest server, by using Cluster WebUI or by running the clpgrp command, the failover group can be started (the operation can be started).

  13. After the failover group is started, make a mirror recovery by clicking Full copy icon in the Mirror disks tab of Cluster WebUI or by executing the following clpmdctrl/clphdctrl command:

    • For mirror disk resources:

      clpmdctrl --recovery <md_resource_name>

    • For hybrid disk resources:

      clphdctrl --recovery <hd_resource_name>

    Note

    Mirror recovery can also be started by using Cluster WebUI or by running the clpmdctrl / clphdctrl command, before starting the failover group.
    In this case, however, the failover group cannot be started unless mirror recovery (full copy) is completed or canceled.

    • For mirror disk resources:

      clpmdctrl --force <copy_source_server> <md_resource_name>

    • For hybrid disk resources:

      clphdctrl --force <copy_source_server> <hd_resource_name>

    See also

    For information on the clpmdctrl / clphdctrl command, see "Reference Guide" -> "EXPRESSCLUSTER command reference" -> "Operating mirror disk resource (clpmdctrl command)" / "Operating hybrid disk resource (clphdctrl command)".

    If forced mirror recovery (the method of not specifying a copy source server) is specified as the --force option, the command is executed on the server that you want to be the latest (status: "Normal" GREEN). After the execution, the failover group can be started (the operation can be started) on the server whose status is "Normal" (GREEN).
    If full copy (the method of specifying a copy source server) is specified as the --force option, the command is executable on any servers. After the execution, mirror recovery (full copy) is started. Once mirror recovery is started, the failover group cannot be started (the operation cannot be started) unless mirror recovery is completed or interrupted.

2.20.3. Restoring the mirror disk on the single server from the disk image

To restore only the mirror disk of the standby server with the active server operating, see "How to restore a virtual machine ~For a mirror disk~", read "the server with the system disk to be restored" as " the server with the mirror disk to be restored", and then follow the procedure from step 1 (moving a failover group) through step 11 (confirming that the mirror is synchronized normally). In steps 4 and 5, create only a virtual hard disk for the mirror disk and replace with it the existing disk.

For the procedure of restoring mirror disks only on the active server or its server group, with the standby server or its server group (not to be restored) stopped, execute restoration as specified in "Simultaneously restoring the mirror disks on both of the active and standby servers from their respective disk images", in which "both" or "both servers" is considered as "single" or "single server", respectively.

Important

  • If you change configuration data (such as a path to a partition) through the procedure, and then the distribution fails for any server, remember to distribute the changed configuration data to the server later.
    If incorrect path information is used, the start of mirroring may fail or the corresponding partition may be destroyed.
  • It is not supported to separately restore, connect, and operate an active server and a standby server according to this procedure.
    There is no problem even in such a case, if mirror recovery (full copy) is executed right after both servers are connected and activated. If operation is carried out without executing mirror recovery (full copy), however, the mirror data can be damaged.

See also

If you want to start the cluster service immediately without waiting for the startup of the other server, run the following command to cancel the cluster activation synchronization wait processing:
clpbwctrl -c
The executed command causes an error if the cluster activation synchronization wait processing is timed out or not yet started.
For information on the clpbwctrl command, see "Reference Guide" -> "EXPRESSCLUSTER command reference" -> "Controlling cluster activation synchronization wait processing (clpbwctrl command)".

2.21. Wait time for synchronized cluster startup

Even all servers in a cluster are powered on simultaneously, it does not always mean that EXPRESSCLUSTER will start up simultaneously on all servers. EXPRESSCLUSTER may not start up simultaneously after rebooting the cluster following shutdown. Because of this, with EXPRESSCLUSTER, if one server is started, it waits for other servers in the cluster to start.

By default, 5 minutes is set to the startup synchronization time. To change the default value, click Cluster Properties in the Cluster WebUI, click Timeout tab, and select Synchronize Wait Time.

For more information, see "Cluster properties Timeout tab" in "Parameter details" in the "Reference Guide".

2.22. Changing disk resources file system

Connect to the Cluster WebUI with a management IP address. If you do not have any management IP address, connect to it by using the actual IP address of any server.

To change the disk resource file system, follow the steps below:

  1. In the operation mode of Cluster WebUI, click Stop Cluster.

  2. Run the following command.
    For example, when the disk resources partition device is /dev/sdb5:
    # clproset -w -d /dev/sdb5
    This makes disk partition of disk resources readable/writable regardless of the EXPRESSCLUSTER behavior.

    Note

    Do not use this command for any other purposes.
    If you use this command when the EXPRESSCLUSTER daemon is active, the file system may be corrupted.

  3. Create the file system in the partition device.

  4. Run the following command to set the disk resources partition to ReadOnly.
    For example, when the disk resources partition device is /dev/sdb5:
    # clproset -o -d /dev/sdb5

  5. Change the configuration data of disk resource file system in the config mode of Cluster WebUI.

  6. Upload the cluster configuration data in the config mode of Cluster WebUI.

  7. In the operation mode of Cluster WebUI, click Start Cluster.

The settings reflecting the changes become effective.

2.23. Changing offset or size of a partition on mirror disk resource

Follow the procedure below when changing the offset (location) or size of the data partition or cluster partition configured on a mirror disk resource after the operation of a cluster is started.

Note

Be sure to follow the steps below to change them. Mirror disk resources may not function properly if you change the partition specified as a data partition or cluster partition only by fdisk.

2.23.1. Data partition configured with LVM

If LVM is used for partitioning, you can extend data partition without re-creating resources or stopping your business (depending on the file system used).

Table 2.1 LVM data partition extension

File system of data partition

Resource re-creation

Business down

Reference

xfs, ext3, ext4, No file system

Not required

Not required

2.23.1.1. Data partition extension for an ext-based or xfs system, or no file system used

Other than above

Required

Required

2.23.1.2. Data partition extension for other file systems

Note

This method is intended for only extension. To shrink partitions, refer to "2.23.2. Data partition configured with other than LVM".

Note

If you follow the instruction below to extend data partition, LVM must be used for the data partition and unused PE (physical extents) of the volume group are sufficient.

2.23.1.1. Data partition extension for an ext-based or xfs system, or no file system used

  1. Confirm the mirror disk resource name you want to resize by [clpstat] command or Cluster WebUI.

  2. For unexpected events, back up partition data in a server where an active group has mirror disk resources you want to resize (use a backup device such as tape device). Note that backup commands to access partition device directly is not supported. Ignore this step if you can discard the data on mirror disk resources.

  3. Confirm the followings:

    • Mirror disk resource status is normal.

    • On both servers, unused PE (physical extents) of the volume group that data partition belongs to are sufficient.

  4. Suspend all the mirror disk monitor resources in the operation mode of Cluster WebUI to prevent automatic mirror recovery.

  5. Run the following [clpmdctrl] command on the server an inactive mirror disk resource belongs to. If the resource is not activated on either server, run the command on either of the servers. The following is an example for extending an md01 data partition to 500 gibibytes.

    # clpmdctrl --resize 500G md01

Important

If a mirror disk resource is activated on either of the servers, make sure to run the command on the server that a deactivated mirror disk belongs to. Execution on an activated server results in a mirror break.

  1. Run the [clpmdctrl] command on the other server. The following is an example for extending an md01 data partition to 500 gibibytes.

    # clpmdctrl --resize 500G md01

  2. If an xfs or ext-based file system is configured on the data partition, extend the file system area by running the command on the server where mirror disk resources are activated.

    <For xfs file systems>
    # xfs_growfs /mnt/nmp1
    Change /mnt/nmp1 as necessary depending on the mirror disk resources mount point.)
    <For ext-based file systems>
    # resize2fs -p /dev/NMP1
    (Replace NMP1 with the mirror partition device name.)

    If you have not configured any file system on the data partition, ignore this step.

  3. In the operation mode of Cluster WebUI, restart all the mirror disk monitor resources that were suspended in step 4.

Important

The [clpmdctrl --resize] command is effective only when mirror disk resources are in the normal status.
If the mirror becomes inconsistent (mirror break) between step 5 and 6, you can not extend a data partition at step 6. In this case, use the [-force] option to forcibly extend the data partition in step 6 and complete all the steps. Then recover the mirror disk.
If you use the [-force] option for extension, full copy is performed to rebuild the mirror first time.
# clpmdctrl --resize -force 500G md01

Note

Data partition size changes depending on PE size.
If PE size is 4M and # clpmdctrl --resize 1022M md01 is specified, the data partition size becomes 1024M and the file system extension limit becomes 1022M.

Note

During the execution of the xfs_growfs command or the resize2fs command , a massive writing process may degrade the operation I/O performance. It is recommended that the execution be performed during off-peak hours.

2.23.1.2. Data partition extension for other file systems

The basic procedure is the same as "2.23.2. Data partition configured with other than LVM".

Note: Use the [lvextend] command instead of [fdisk] to resize partition size.

2.23.2. Data partition configured with other than LVM

2.23.2.1. When not changing a device name of a partition on mirror disk resource

  1. Check the name of a mirror disk resource whose size you want to change by the clpstat command or by the Cluster WebUI.

  2. On the server where a group with a mirror disk resource whose size you want to change is activated, back up the data in a partition to a device such as tape. Note that backup commands that access a partition device directly are not supported.
    This step is not required if there is no problem to discard the data on a mirror disk resource.
    Server 1 and Server 2 with different disks connected respectively

    Fig. 2.26 Mirror disk resources activated on Server 1

    Server 1 with a disk and a backup device connected, and Server 2 with a disk connected

    Fig. 2.27 Back up the data on Server 1

  3. Set the EXPRESSCLUSTER service not to start up on either of the servers.

    clpsvcctrl.sh --disable core
    Server 1 and Server 2 with different disks connected respectively

    Fig. 2.28 Set the EXPRESSCLUSTER service not to start

  4. Shut down a cluster, and then restart the OS.
    To shut down a cluster, run the clpstdn command on either of a server, or execute a cluster shutdown on the Cluster WebUI.
    Server 1 and Server 2 with different disks connected respectively

    Fig. 2.29 Execute a cluster shutdown on either of the servers

    Server 1 and Server 2 with different disks connected respectively

    Fig. 2.30 Restart the OS

  5. On both servers, run the fdisk command to change the offset or size of a partition.

    Server 1 and Server 2 with different disks connected respectively

    Fig. 2.31 Change the size of a partition

  6. Run the following command on both servers.

    # clpmdinit --create force <Mirror_disk_resource_name>
    Server 1 and Server 2 with different disks connected respectively

    Fig. 2.32 Initialize a cluster partition

    Server 1 and Server 2 with different disks connected respectively

    Fig. 2.33 Execute the first mkfs to create a file system

    Note

    When you set Execute initial mkfs to off in the mirror disk resource setting, mkfs will not be executed automatically. Please execute mkfs manually to the data partition of mirror disk resource.

  7. Set the EXPRESSCLUSTER service to start up on both servers.

    clpsvcctrl.sh --enable core
    Server 1 and Server 2 with different disks connected respectively

    Fig. 2.34 Set the EXPRESSCLUSTER service to start

  8. Run the reboot command to restart both servers. The servers are started as a cluster.

  9. After a cluster is started, the same process as the initial mirror construction at cluster creation is performed. Run the following command or use the Cluster WebUI to check if the initial mirror construction is completed.

    # clpmdstat --mirror <Mirror_disk_resource_name>
    Server 1 and Server 2 with different disks connected respectively

    Fig. 2.35 Start mirror recovery on Server 1

  10. When the initial mirror construction is completed and a failover group starts, a mirror disk resource becomes active.

    Server 1 and Server 2 with different disks connected respectively

    Fig. 2.36 Initial mirror construction is completed

  11. On the server where a group with a mirror partition whose size you changed is activated, restore the data you backed up. Note that backup commands that access a partition device directly are not supported.
    This step is not required if there is no problem to discard the data on a mirror disk resource.
    Server 1 with a disk and a backup device connected, and Server 2 with a disk connected

    Fig. 2.37 Restore the data that was backed up

2.23.2.2. When changing a device name of a partition on mirror disk resource

  1. Check the name of a mirror disk resource whose size you want to change by the clpstat command or by the Cluster WebUI.

  2. On the server where a group with a mirror disk resource whose size you want to change is activated, back up the data in a partition to a device such as tape. Note that backup commands that access a partition device directly are not supported.
    This step is not required if destroying the data on a mirror disk resource does not cause any problem.
    Server 1 and Server 2 with different disks connected respectively

    Fig. 2.38 Mirror disk resources activated on Server 1

    Server 1 with a disk and a backup device connected, and Server 2 with a disk connected

    Fig. 2.39 Back up the data on Server 1

  3. Set the EXPRESSCLUSTER service not to start up on either of the servers.

    clpsvcctrl.sh --disable core
    Server 1 and Server 2 with different disks connected respectively

    Fig. 2.40 Set the EXPRESSCLUSTER service not to start

  4. Shut down a cluster, and then restart the OS.
    To shut down a cluster, run the clpstdn command on either of a server, or execute a cluster shutdown on the WebManager.
    Server 1 and Server 2 with different disks connected respectively

    Fig. 2.41 Execute a cluster shutdown on either of the servers

    Server 1 and Server 2 with different disks connected respectively

    Fig. 2.42 Restart the OS

  5. On both servers, run the fdisk command to change the offset or size of a partition.

    Server 1 and Server 2 with different disks connected respectively

    Fig. 2.43 Change the size of a partition

  6. Change and upload the cluster configuration data. Change a mirror disk resource as described in "Modifying the cluster configuration data by using the Cluster WebUI" in "Modifying the cluster configuration data" in the "Installation and Configuration Guide".

  7. Run the following command on the both servers.

    # clpmdinit --create force <Mirror_disk_rseource_name>
    Server 1 and Server 2 with different disks connected respectively

    Fig. 2.44 Initialize a cluster partition

    Server 1 and Server 2 with different disks connected respectively

    Fig. 2.45 Execute the first mkfs to create a file system

    Note

    When you set Execute initial mkfs to off in the mirror disk resource setting, mkfs will not be executed automatically. Please execute mkfs manually to the data partition of mirror disk resource.

  8. Set the EXPRESSCLUSTER service to start up on both servers.

    clpsvcctrl.sh --enable core
    Server 1 and Server 2 with different disks connected respectively

    Fig. 2.46 Set the EXPRESSCLUSTER service to start

  9. Run the reboot command to restart both servers. The servers are started as a cluster.

  10. After a cluster is started, the same process as the initial mirror construction at cluster creation is performed. Run the following command or use the Cluster WebUI to check if the initial mirror construction is completed.

    # clpmdstat --mirror <Mirror_disk_resource_name>
    Server 1 and Server 2 with different disks connected respectively

    Fig. 2.47 Start mirror recovery on Server 1

  11. When the initial mirror construction is completed and a failover group starts, a mirror disk resource becomes active.

    Server 1 and Server 2 with different disks connected respectively

    Fig. 2.48 Initial mirror construction is completed

  12. On the server where a group with a mirror partition whose size you changed is activated, restore the data you backed up. Note that backup commands that access a partition device directly are not supported.
    This step is not required if there is no problem to discard the data on a mirror disk resource.
    Server 1 with a disk and a backup device connected, and Server 2 with a disk connected

    Fig. 2.49 Restore the data that was backed up

2.24. Changing offset or size of a partition on hybrid disk resource

Follow the procedure below when changing the offset (location) or size of the data partition or cluster partition configured on a hybrid disk resource after the operation of a cluster is started.

Note

Be sure to follow the steps below to change them. Hybrid disk resources may not function properly if you change the partition specified as a data partition or cluster partition only by fdisk.

2.24.1. With the data partition configured with LVM

If LVM is used for partitioning, you can extend the data partition without re-creating resources or stopping your business (depending on the file system used).

Table 2.2 LVM data partition extension

File system of data partition

Resource re-creation

Business down

Reference

xfs, ext3, ext4, or no file system (none)

Not required

Not required

2.24.1.1. Expanding data partition for an ext-based or xfs system, or no file system used

Other than above

Required

Required

2.24.1.2. Expanding data partition for other file systems

Note

This method is intended only for extension. To shrink partitions, refer to "2.24.2. With the data partition configured without LVM".

Note

If you follow the instruction below to extend data partition, LVM must be used for the data partition, and unused PE (physical extents) of the volume group must be sufficient.

2.24.1.1. Expanding data partition for an ext-based or xfs system, or no file system used

  1. Run the [clpstat] command or use Cluster WebUI to confirm the name of a hybrid disk resource you want to resize.

  2. On the server containing the activated group with the hybrid disk resource to be resized, back up the data in a partition to a device (such as tape) for unexpected events. However, there is no support for any backup command for direct access to the partition device. You can skip this step if there is no problem with discarding the data on the hybrid disk resource.

  3. Confirm the following:

    • The status of the hybrid disk resource is normal.

    • On both servers, there is sufficient, unused PE (physical extent) of the volume group that the data partition belongs to.

  4. Suspend all the hybrid disk monitor resources in the operation mode of Cluster WebUI to prevent automatic mirror recovery.

  5. Keeping in operation the current server of each server group, shut down all the other servers. You can check the status of the current server by executing the clphdstat with the -a option. The following shows an example of checking the status of the current server in the hd01 resource:

    clphdstat -a hd01
  6. Execute the following clphdctrl command on the current server of the server group where the hybrid disk resource is deactivated.
    If the resource is not activated on either server group, run the command on either of the servers. The following is an example for extending an hd01 data partition to 500 gibibytes.
    # clphdctrl --resize 500G hd01

Important

If the hybrid disk resource is activated on either of the servers, make sure to run the command on the server where the hybrid disk resource is deactivated. Execution on an active server group results in a mirror break.

  1. Likewise, perform the following clphdctrl command on the current server of the other server group.
    The following is an example for extending an hd01 data partition to 500 gibibytes.
    # clphdctrl --resize 500G hd01

  2. If an xfs or ext-based file system is configured on the data partition, extend the file system area by running the command on the server where hybrid disk resources are activated.

    <For xfs file systems>
    # xfs_growfs /mnt/nmp1
    (Change /mnt/nmp1 as necessary depending on the hybrid disk resources mount point.)
    <For ext-based file systems>
    # resize2fs -p /dev/NMP1
    (Replace NMP1 with the mirror partition device name.)

    You can skip this step if no file system is used for the data partition (none).

  3. In the operation mode of Cluster WebUI, restart all the hybrid disk monitor resources that were suspended in step 4.

  4. Start up all the servers that you shut down in step 5.

Important

The [clpmdctrl --resize] command is effective only when hybrid disk resources are in the normal status.
If the mirror becomes inconsistent (mirror break) between step 5 and 6, the data partition cannot be extended at step 6. In this case, use the -force option to forcibly extend the data partition in step 6 and complete all the steps. Then recover the mirror disk.
If the [-force] option is used for extension, full copy is performed to rebuild the mirror first time.
# clphdctrl --resize -force 500G hd01

Note

The size of the data partition depends on that of PE.
If the size of PE is 4M and # clphdctrl --resize 1022M hd01 is specified, the size of the data partition becomes 1024M and the limit of the file system extension becomes 1022M.

Note

During the execution of xfs_growfs and resize2fs, a massive writing process may degrade the operation I/O performance. It is recommended that the execution be performed during off-peak hours.

2.24.1.2. Expanding data partition for other file systems

The basic procedure is the same as "2.24.2. With the data partition configured without LVM".

Note: Use the [lvextend] command instead of [fdisk] to resize a partition size.

2.24.2. With the data partition configured without LVM

2.24.2.1. When not changing a device name of a partition on hybrid disk resource

  1. Check the name of a hybrid disk resource whose size you want to change by the clpstat command or by the Cluster WebUI.

  2. On the server where a group with the hybrid disk resource whose size you want to change is activated, back up the data in a partition to a device such as tape. Note that backup commands that access a partition device directly are not supported.
    This step is not required if there is no problem to discard the data on the hybrid disk resource.
    Server 1 and Server 2 both with the same shared disk connected, and Server 3 with a disk connected

    Fig. 2.50 Server 1 containing the activated group with the hybrid disk resource

    Server 1 with a shared disk and a backup device connected, Server 2 with the same shared disk connected, and Server 3 with a disk connected

    Fig. 2.51 Back up the data on Server 1

  3. Set the EXPRESSCLUSTER service not to start up on any of all the servers.

    clpsvcctrl.sh --disable core
    Server 1 and Server 2 both with the same shared disk connected, and Server 3 with a disk connected

    Fig. 2.52 Set the EXPRESSCLUSTER service not to start

  4. Shut down a cluster, and then restart the OS.
    To shut down a cluster, run the clpstdn command on either of a server, or execute a cluster shutdown on the Cluster WebUI.
    Server 1 and Server 2 both with the same shared disk connected, and Server 3 with a disk connected

    Fig. 2.53 Execute a cluster shutdown on either of the servers

    Server 1 and Server 2 both with the same shared disk connected, and Server 3 with a disk connected

    Fig. 2.54 Restart the OS

  5. Run the fdisk command on a server to change the offset or size of a partition. When servers are connected to the shared disk, run the fdisk from either of the servers for the change.

    Server 1 and Server 2 both with the same shared disk connected, and Server 3 with a disk connected

    Fig. 2.55 Change the size of a partition

  6. Run the following command on a server. When servers are connected to the shared disk, run the command on the server where the command in previous step was executed.
    # clphdinit --create force <Hybrid disk resource name>
    Server 1 and Server 2 both with the same shared disk connected, and Server 3 with a disk connected

    Fig. 2.56 Initialize a cluster partition

  7. Run the following command on a server.When servers are connected to the shared disk, run the command on the server where the command in previous step was executed.
    # mkfs -t <Type of Filesystem>* <Data Partition>
    Server 1 and Server 2 both with the same shared disk connected, and Server 3 with a disk connected

    Fig. 2.57 Execute the first mkfs to create a file system

  8. Set the EXPRESSCLUSTER service to start up on all servers.

    clpsvcctrl.sh --enable core
    Server 1 and Server 2 both with the same shared disk connected, and Server 3 with a disk connected

    Fig. 2.58 Set the EXPRESSCLUSTER service to start

  9. Run the reboot command to restart all servers. The servers are started as a cluster.

  10. After the cluster is started, the same process as the initial mirror construction at cluster creation is performed. Run the following command or use the Cluster WebUI to check if the initial mirror construction is completed.

    # clphdstat --mirror <hybrid_disk_resource_name>
    Server 1 and Server 2 both with the same shared disk connected, and Server 3 with a disk connected

    Fig. 2.59 Start mirror recovery on Server 1

  11. When the initial mirror construction is completed and a failover group starts, a hybrid disk resource becomes active.

    Server 1 and Server 2 both with the same shared disk connected, and Server 3 with a disk connected

    Fig. 2.60 Initial mirror construction is completed

  12. On the server where a group with the partition whose size you changed is activated, restore the data you backed up. Note that backup commands that access a partition device directly are not supported.
    This step is not required if there is no problem to discard the data on a hybrid disk resource.
    Server 1 with a shared disk and a backup device connected, Server 2 with the same shared disk connected, and Server 3 with a disk connected

    Fig. 2.61 Restore the data that was backed up

2.24.2.2. When changing a device name of a partition on hybrid resource

  1. Check the name of a hybrid disk resource whose size you want to change by the clpstat command or by the Cluster WebUI.

  2. On the server where a group with the hybrid disk resource whose size you want to change is activated, back up the data in a partition to a device such as tape. Note that backup commands that access a partition device directly are not supported.
    This step is not required if destroying the data on the hybrid disk resource does not cause any problem.
Server 1 and Server 2 both with the same shared disk connected, and Server 3 with a disk connected

Fig. 2.62 Server 1 containing the activated group with the hybrid disk resource

Server 1 with a shared disk and a backup device connected, Server 2 with the same shared disk connected, and Server 3 with a disk connected

Fig. 2.63 Back up the data on Server 1

  1. Set the EXPRESSCLUSTER service not to start up on any of all the servers.

    clpsvcctrl.sh --disable core
    Server 1 and Server 2 both with the same shared disk connected, and Server 3 with a disk connected

    Fig. 2.64 Set the EXPRESSCLUSTER service not to start

  2. Shut down a cluster, and then restart the OS.
    To shut down a cluster, run the clpstdn command on either of a server, or execute a cluster shutdown on the Cluster WebUI.
    Server 1 and Server 2 both with the same shared disk connected, and Server 3 with a disk connected

    Fig. 2.65 Execute a cluster shutdown on either of the servers

    Server 1 and Server 2 both with the same shared disk connected, and Server 3 with a disk connected

    Fig. 2.66 Restart the OS

  3. On a server, run the fdisk command to change the offset or size of a partition. When servers are connected to the shared disk, run the fdisk command from either of servers to change.

    Server 1 and Server 2 both with the same shared disk connected, and Server 3 with a disk connected

    Fig. 2.67 Change the size of a partition

  4. Change and upload the cluster configuration data. Change a hybrid disk resource as described in "Modifying the cluster configuration data by using the Cluster WebUI" in "Modifying the cluster configuration data" in the "Installation and Configuration Guide".

  5. Run the following command on the server. When servers are connected to the shared disk, execute the command on the server where the command was executed in step 5.

    # clphdinit --create force <Hybrid_disk_reseource_name>
    Server 1 and Server 2 both with the same shared disk connected, and Server 3 with a disk connected

    Fig. 2.68 Initialize a cluster partition

  6. Run the following command on the server.When servers are connected to the shared disk, run the command on the server where the command in previous step was executed.
    # mkfs -t <Type of Filesystem> <Data Partition>
    Server 1 and Server 2 both with the same shared disk connected, and Server 3 with a disk connected

    Fig. 2.69 Execute the first mkfs to create a file system

  7. Set the EXPRESSCLUSTER service to start up on all servers.

    clpsvcctrl.sh --enable core
    Server 1 and Server 2 both with the same shared disk connected, and Server 3 with a disk connected

    Fig. 2.70 Set the EXPRESSCLUSTER service to start

  8. Run the reboot command to restart all servers. The servers are started as a cluster.

  9. After the cluster is started, the same process as the initial mirror construction at cluster creation is performed. Run the following command or use the Cluster WebUI to check if the initial mirror construction is completed.

    # clphdstat --mirror <Hybrid_disk_resource_name>
    Server 1 and Server 2 both with the same shared disk connected, and Server 3 with a disk connected

    Fig. 2.71 Start mirror recovery on Server 1

  10. When the initial mirror construction is completed and a failover group starts, a hybrid disk resource becomes active.

    Server 1 and Server 2 both with the same shared disk connected, and Server 3 with a disk connected

    Fig. 2.72 Initial mirror construction is completed

  11. On the server where a group with the partition whose size you changed is activated, restore the data you backed up. Note that backup commands that access a partition device directly are not supported.
    This step is not required if there is no problem to discard the data on the hybrid disk resource.
    Server 1 with a shared disk and a backup device connected, Server 2 with the same shared disk connected, and Server 3 with a disk connected

    Fig. 2.73 Restore the data that was backed up

2.25. Changing the server configuration (add/delete)

2.25.1. Adding a server (mirror disk or hybrid disk is not used)

To add a server, follow the steps below:

Important

  1. Make sure that the cluster is working normally.

  2. Install the EXPRESSCLUSTER Server on a new server. For details, see "Installing the EXPRESSCLUSTER RPM" in "Setting up the EXPRESSCLUSTER Server" in "Installing EXPRESSCLUSTER" in the "Installation and Configuration Guide". Restart the server on which the EXPRESSCLUSTER Server was installed.

  3. Access the other server in the cluster with a Web browser and click Add server in the Cluster WebUI config mode.

  4. By using the config mode of Cluster WebUI, configure the following settings of the Add server.

    • Information on the Source IP Address of the server to add, on the Details tab of Properties of the virtual IP resource (when using the virtual IP resource).

    • Information on the ENI ID of the server to add, on the Details tab of Properties of the AWS elastic IP resource (when using the AWS elastic IP resource).

    • Information on the ENI ID of the server to add, on the Details tab of Properties of the AWS virtual IP resource (when using the AWS virtual IP resource).

    • Information on the ENI ID of the server to add, on the Details tab of Properties of the AWS secondary IP resource (when using the AWS secondary IP resource).

    • Information on the IP Address of the server to add, on the Details tab of Properties of the Azure DNS resource (when using the Azure DNS resource).

    • Information on the IP Address of the server to add, on the Details tab of Properties of the Google Cloud DNS resource (when using the Google Cloud DNS resource).

    • Information on the Region, Zone OCID, and IP Address of the server to add, on the Details tab of Properties of the Oracle Cloud DNS resource (when using the Oracle Cloud DNS resource).

  5. Click Apply the Configuration File in the config mode of Cluster WebUI to apply the cluster configuration information on the cluster.

    Note: Apply the configuration when the confirmation message is displayed.

  6. Perform Start server service of the server added from the Cluster WebUI config mode.

  7. Click Refresh data in the operation mode of Cluster WebUI to verify the cluster is properly working.

2.25.2. Adding a server (Mirror disk or hybrid disk is used)

To add a server, follow the steps below:

Important

  1. Make sure that the cluster is working normally.

  2. Install the EXPRESSCLUSTER Server on a new server. For details, see "Installing the EXPRESSCLUSTER RPM" in "Setting up the EXPRESSCLUSTER Server" in "Installing EXPRESSCLUSTER" in the "Installation and Configuration Guide". Restart the server on which the EXPRESSCLUSTER Server was installed.

  3. In the operation mode of Cluster WebUI, click Stop cluster.

  4. Perform Stop Mirror Agent in the Cluster WebUI operation mode.

  5. Access to other server in the cluster via the Web browser and click the server to add in the config mode of Cluster WebUI.

  6. By using the config mode of Cluster WebUI, configure the following settings of the Add server.

    • Information on the Source IP Address of the server to add, on the Details tab of Properties of the virtual IP resource (when using the virtual IP resource).

    • Information on the ENI ID of the server to add, on the Details tab of Properties of the AWS elastic IP resource (when using the AWS elastic IP resource).

    • Information on the ENI ID of the server to add, on the Details tab of Properties of the AWS virtual IP resource (when using the AWS virtual IP resource).

    • Information on the ENI ID of the server to add, on the Details tab of Properties of the AWS secondary IP resource (when using the AWS secondary IP resource).

    • Information on the IP Address of the server to add, on the Details tab of Properties of the Azure DNS resource (when using the Azure DNS resource).

    • Information on the IP Address of the server to add, on the Details tab of Properties of the Google Cloud DNS resource (when using the Google Cloud DNS resource).

    • Information on the Region, Zone OCID, and IP Address of the server to add, on the Details tab of Properties of the Oracle Cloud DNS resource (when using the Oracle Cloud DNS resource).

  7. When using a hybrid disk resource in the added server, click Properties of Servers in the Conf mode of Cluster WebUI. From the Server Group tab, add the server to the servers that can run the Group. Do this for required servers only.

  8. Click Apply the Configuration File in the config mode of Cluster WebUI to apply the cluster configuration information on the cluster. Select OK when the service restart dialog appears.

  9. Perform Start Mirror Agent in the Cluster WebUI operation mode.

  10. In the operation mode of Cluster WebUI, click Start cluster.

  11. Click Refresh data in the operation mode of Cluster WebUI to verify the cluster is properly working.

2.25.3. Deleting a server (Mirror disk or hybrid disk is not used)

To delete a server, follow the steps below:

Important

  • When adding a server in changing the cluster configuration, do not make any other changes such as adding a group resource.

  • Refer to the following information for licenses registered in the server you want to delete.

    • No action required for CPU licenses.

    • VM node licenses and other node licenses are discarded when EXPRESSCLUSTER is uninstalled.
      Back up the serial numbers and keys of licenses if required.

    • No action required for fixed term licenses. Unused licenses are automatically collected and provided to other servers.

  1. Make sure that the cluster is working normally. If any group is active on the server you are going to delete, move the group to another server.

  2. When the server to be deleted is registered in a server group, click Properties of Server of the config mode of Cluster WebUI. Delete the server from Servers that can run the Group in the Server Group tab.

  3. Click Remove Server of the server to delete in the config mode of Cluster WebUI.

  4. Click Apply the Configuration File in the config mode of Cluster WebUI to apply the cluster configuration information on the cluster.

    Note: Apply the configuration when the confirmation message is displayed.

  5. Click Refresh data in the operation mode of Cluster WebUI to verify the cluster is properly working.

  6. Deleted servers will not belong to clusters. To uninstall EXPRESSCLUSTER servers, refer to "Installation and Configuration Guide" > "Uninstalling and reinstalling EXPRESSCLUSTER" > "Uninstallation" > " Uninstalling the EXPRESSCLUSTER Server".
    Note: In the uninstallation reference, assume server rebooting as OS rebooting in the deleted server.

2.25.4. Deleting a server (Mirror disk or hybrid disk is used)

To delete a server, follow the steps below:

Important

  • When deleting a server in changing the cluster configuration, do not make changes (such as adding a group resource) other than ones given below.

  • Refer to the following information for licenses registered in the server you want to delete.

    • No action required for CPU licenses.

    • VM node licenses and other node licenses are discarded when EXPRESSCLUSTER is uninstalled.
      Back up the serial numbers and keys of licenses if required.

    • No action required for fixed term licenses. Unused licenses are automatically collected and provided to other servers.

  1. Make sure that the cluster is working normally. If any group is active on the server you are going to delete, move the group to another server.

  2. In the operation mode of Cluster WebUI, click Stop cluster.

  3. Perform Stop Mirror Agent in the Cluster WebUI operation mode.

  4. Click Remove resource of mirror disk resources or hybrid disk resources in the Cluster WebUI config mode.

  5. When the server to be deleted is registered in a server group, click Properties of Server of the config mode of Cluster WebUI. Delete the server from Servers that can run the Group in the Server Group tab.

  6. Click Remove Server of the server to delete in the config mode of Cluster WebUI.

  7. Click Apply the Configuration File in the config mode of Cluster WebUI to apply the cluster configuration information on the cluster.

  8. In the operation mode of Cluster WebUI, click Start Mirror Agent (if Mirror Agent is stopped) and then Start Cluster. Perform Start Mirror Agent and Start cluster in the Cluster WebUI operation mode.

  9. Click Refresh data in the operation mode of Cluster WebUI to verify the cluster is properly working.

  10. Deleted servers will not belong to clusters. To uninstall EXPRESSCLUSTER servers, refer to "Installation and Configuration Guide" > "Uninstalling and reinstalling EXPRESSCLUSTER" > "Uninstallation" > "Uninstalling the EXPRESSCLUSTER Server".
    Note: In the uninstallation reference, assume server rebooting as OS rebooting in the deleted server.

2.26. Changing the server IP address

To change the server IP address after you have started the cluster system operation, follow the instructions below.

2.26.1. Changing the interconnect IP address / mirror disk connect IP address

  1. Use the clpstat command or the Cluster WebUI to verify all servers in the cluster are working normally.

  2. Back up the cluster configuration data. Use the clpcfctrl command to back up the data.
    If you have the configuration data that contains the data at the cluster creation, use that configuration data.

  3. In the config mode of Cluster WebUI, change the server IP address based on the back up cluster configuration data, and then save it.

  4. Disable the startup settings of the EXPRESSCLUSTER daemon in all servers in the cluster. For more information, see "Disabling the EXPRESSCLUSTER daemon"in "Suspending EXPRESSCLUSTER" in "Preparing to operate a cluster system" in the Installation and Configuration Guide.

  5. By the clpstdn command or in the operation mode of Cluster WebUI, to shut down the cluster, and then restart all servers.

  6. Change the IP address. If a server reboot is required after changing the IP address, run the reboot command or use other means on the server where the IP address has changed.

  7. Verify the changed IP address is valid by running the ping command or using other means.

  8. Distribute the cluster configuration data to all the servers. Use the clpcfctrl command to deliver the data.

  9. Enable the startup settings of the EXPRESSCLUSTER daemon in all servers in the cluster.

  10. Run the reboot command or use other means on all servers in the cluster to reboot them.

  11. Use the clpstat command or the Cluster WebUI to verify all servers in the cluster are working normally.

2.26.2. Changing only the subnet mask of the interconnect IP address

  1. Use the clpstat command or the Cluster WebUI to verify all servers in the cluster are working normally.

  2. Back up the cluster configuration data. Use the clpcfctrl command to back up the data.
    If you have the configuration data that contains the data at the cluster creation, use that configuration data.

  3. In the config mode of Cluster WebUI, change the server IP address based on the back up cluster configuration data, and then save it.

  4. Disable startup settings of the EXPRESSCLUSTER daemon in all servers in the cluster.

  5. By the clpstdn command or in the operation mode of Cluster WebUI, to shut down the cluster, and then restart all servers.

  6. Change the subnet mask of the IP address. If server reboot is required after changing the subnet mask of IP address, run the reboot command or use other means on the server where the subnet mask of the IP address has been changed.

  7. Verify the changed IP address is valid by running the ping command or using other means.

  8. Distribute the cluster configuration data to all servers. Use the clpcfctrl command to deliver the data.

  9. Enable the startup settings of the EXPRESSCLUSTER daemon in all servers in the cluster.

  10. Run the reboot command or use other means on all the servers in the cluster.

  11. Use the clpstat command or the Cluster WebUI to verify all the servers in the cluster are working normally.

2.27. Changing the host name

Follow the steps below if you want to change the host name of a server after you have started the cluster system operation.

2.27.1. Changing the host name

  1. Use the clpstat command or the Cluster WebUI to verify all the servers in the cluster are working normally.

  2. Back up the cluster configuration data. Use the clpcfctrl command to back up the data.
    If you have the configuration data that contains the data at the cluster creation, use that configuration data

  3. In the config mode of Cluster WebUI, change the host name of your target server based on the back up cluster configuration data, and then save it.

  4. Disable the startup settings of the EXPRESSCLUSTER daemon in all servers in the cluster. For more information, see "Disabling the EXPRESSCLUSTER daemon" in "Suspending EXPRESSCLUSTER in "Preparing to operate a cluster system" in the "Installation and Configuration Guide".

  5. By the clpstdn command or in the operation mode of Cluster WebUI, to shut down the cluster, and then restart all servers.

  6. Change the host name. If the server needs to be rebooted after changing the host name, run the reboot command or use other means on the server.

  7. Verify the changed host name is valid by running the ping command or using other means.

  8. Distribute the cluster configuration data to all the servers. Use the clpcfctrl command to deliver the data. Executing the clpcfctrl command requires the --nocheck option.

    Note

    Check cluster configuration information before the distribution if required.

  9. Enable the startup settings of the EXPRESSCLUSTER daemon in all servers in the cluster.

  10. Run the reboot command or use other means on all the servers in the cluster to reboot them.

  11. Use the clpstat command or the Cluster WebUI to verify all the servers in the cluster are in the normal status.

See also

For information on troubleshooting clpcfctrl problems, see "Changing, backing up, and checking cluster configuration data (clpcfctrl command)" in "EXPRESSCLUSTER command reference" in the "Reference Guide".
For details on how to stop and start daemons, see "Suspending EXPRESSCLUSTER" in "Preparing to operate a cluster system" in the "Installation and Configuration Guide".

2.28. How to add a resource without stopping the group

You can add, to a group that is already running, a resource that supports dynamic resource addition without stopping the group.

Group resources that currently support dynamic resource addition are as follows:

Group resource name

Abbreviation

Supported version

Exec resource

exec

4.0.0-1 or later

Disk resource

disk

4.0.0-1 or later

Floating IP resource

fip

4.0.0-1 or later

Virtual IP resource

vip

4.0.0-1 or later

Volume manager resource

volmgr

4.0.0-1 or later

See also

If all the resources in the group to which the resource to add will belong have been started normally, the resource to add will also be started.
If at least one of the resources in the group to which the resource to add will belong is in the activation or deactivation error state, the dynamic resource addition function will be disabled and group stoppage will be requested. If the group is in the stopped state, the resource will be added and placed in the stopped state.

Perform the following procedure to dynamically add a resource after starting the operation.

2.28.1. How to dynamically add a resource

  1. Confirm that all servers in the cluster are operating normally by running the [clpstat] command or using the Cluster WebUI.

  2. Confirm that all resources in the group to which a resource is added are started normally by running the [clpstat] command or using the Cluster WebUI.

  3. Use the config mode of Cluster WebUI to add a resource to the group and save it.

  4. Run the [clpcl --suspend] command or use the operation mode of Cluster WebUI to suspend the cluster.

  5. Distribute the cluster configuration data to all the servers. Run the [clpcfctrl] command to deliver the data. Run the following command to dynamically add a resource.

    Do either of the following depending on the type of configuration data saved in the config mode of Cluster WebUI.

    clpcfctrl --dpush -x <path of configuration data file>

  6. Run the [clpcl --resume] command or use the operation mode of Cluster WebUI to resume the cluster.

  7. Confirm that the resource has been added by running the [clpstat] command or using the Cluster WebUI.

See also

For information on troubleshooting [clpcfctrl] problems, see "Changing, backing up, and checking cluster configuration data (clpcfctrl command)" in "EXPRESSCLUSTER command reference" in the "Reference Guide".

2.29. Updating data encryption key file of mirror/hybrid disk resources

Perform the following procedure to update the encryption key used for the mirror communication encryption of mirror disk resources/hybrid disk resources.

Note

The following procedure is executable while mirror disk resources and hybrid disk resources are activated. At this time, however, mirroring in progress is suspended. In this case, execute mirror recovery after the completion of the procedure.

  1. Run the openssl command to create a new encryption key file:

    openssl rand 32 -out newkeyfile.bin

  2. Overwrite the encryption key files for all the servers of which mirror disk resources/hybrid disk resources can be activated, by using the file created at step 1. Keep the original file then.

  3. Execute the --updatekey option for the clpmdctrl or clphdctrl command.

    • for mirror disk resources

      clpmdctrl --updatekey md01

    • for hybrid disk resources

      clphdctrl --updatekey hd01
    Once you execute the option on either server on which resources can be activated, the key information is updated for all servers necessary for updated.
    At this time, mirroring in progress is suspended.

  4. Updating of encryption key information is completed. From now on, the mirror communication encryption/decryption is executed by using the new encryption key.

  5. If necessary, perform mirror recovery to resume the suspended mirroring.