Hitachi

JP1 Version 12 JP1/Integrated Management 2 - Manager Overview and System Design Guide


13.7.11 Considering the timeout period during startup or stop of JP1/IM - Manager (in UNIX)

JP1/IM - Manager performs operations according to the specified timeout period at its startup or stop. If it takes long for the startup or stop operation and the default timeout period expires, JP1/IM - Manager times out before the startup or stop operation finishes.

You can handle such a case by changing the default timeout period at startup or stop.

The default timeout periods are as follows:

#

This value is obtained by summing up the following timeout values, because JP1/IM - Manager is stopped and then the IM database is stopped:

- JP1/IM - Manager: 240 seconds

- IM database: 60 seconds

Organization of this subsection

(1) Guidelines for the timeout values

Different guidelines for the timeout values are suggested depending on the size of the IM database. The following table lists the guidelines for timeout values at startup and stop as well as the default values of the jco_start and jco_stop commands on an IM-database-size basis.

Table 13‒13: Guidelines for the startup and stop timeout values

Size

Startup timeout period (seconds)

Stop timeout period (seconds)#

Small-scale (S)

120

120

Medium-scale (M)

180

150

Large-scale (L)

300 (Default)

240 (Default)

#

This value does not contain the timeout value (60 seconds) of the IM database.

Important
  • The startup period of the IM database is not considered at startup because the database is started in the background.

  • Specify the timeout value at stop with the timeout period of the IM database, 60 seconds, in mind because the database is stopped after JP1/IM - Manager is stopped.

  • The startup and stop times of machines depend on their performance and status. Therefore, a timeout might occur even if the guideline values are specified. Use proper timeout values according to your operating environment.

    For details about how to calculate the operation time for startup or stop from log information at startup or stop of JP1/IM - Manager, see 13.7.11(2) How to measure the startup or stop time from log information.

  • In cluster use, review any timeout values for the cluster software.

For details about the configuration of the IM database, see 13.1.3 Estimating IM database capacity requirements.

(2) How to measure the startup or stop time from log information

In the integrated trace log in the operating environment, check the times of the message IDs at startup and stop of JP1/IM - Manager to calculate the difference between the start and end times. You can check the processing time at startup or stop.

(a) How to calculate the startup time

To configure the jco_start command, follow these steps:

  1. In the integrated trace log, check the times of the following message IDs:

    Start time of startup

    KAVB3705-I: evflow is now starting... pid = process-ID

    End time of startup

    KAVB3603-I: All processes of JP1_CONS are running.

  2. Calculate the startup period of JP1/IM - Manager by subtracting the start time of startup from the end time of startup.

    Example of the start time of startup and the end time of startup

    0031 2020/07/01 14:11:46.750     JCO_SPMD         00004642 -------- KAVB3705-I evflow is now starting... pid = process-ID
    0046 2020/07/01 14:12:31.794     JCO_SPMD         00004642 -------- KAVB3603-I All processes of JP1_CONS are running.

    Calculation result of the above

    14:12:31.794 - 14:11:46.750 = About 45 seconds
  3. Specify the timeout period for startup, based on the calculated time.

    Based on the calculated value, specify a value large enough to complete the operation with the load on the system in mind.

(b) Configure jco_stop

To calculate the stop time, follow these steps:

  1. In the integrated trace log, check the times of the following message IDs:

    Start time of stop

    KAVB3577-I : jco_spmd_stop : A stop request will now be sent to all processes of JP1_CONS.

    End time of stop

    KAVB3743-I: JP1_CONS has terminated.

  2. Calculate the stop period of JP1/IM - Manager by subtracting the start time of stop from the end time of stop.

    Example of the start time of stop and the end time of stop

    0041 2020/07/01 15:20:01.061     JCO_SPMD         0000485E -------- KAVB3577-I jco_spmd_stop : A stop request will now be sent to all processes of JP1_CONS.
    0059 2020/07/01 15:20:34.715     JCO_SPMD         0000485E -------- KAVB3743-I JP1_CONS has terminated.

    Calculation result of the above

    15:20:34.715 - 15:20:01.061 = About 34 seconds
  3. Specify the timeout period for stop, based on the calculated time.

    Based on the calculated value, specify a value large enough to complete the operation with the load on the system in mind.

(3) Setting the timeout value for startup or stop of JP1/IM - Manager

The following describes how to set the timeout value for startup or stop of JP1/IM - Manager.

After changing the timeout value, you can start or stop JP1/IM - Manager with the similar method for normal startup or stop. For details about startup and stop, see Chapter 3. Starting and Stopping JP1/IM - Manager in the JP1/Integrated Management 2 - Manager Administration Guide.

(a) Configure jco_start

To configure the jco_start command, follow these steps:

  1. Copy jco_start.model to a file with a given name.

    cd /etc/opt/jp1cons
    cp -p jco_start.model given-name
  2. Open and edit the copied script in a text editor.

    To change the timeout value, change the value of MAX_RETRY#1. For details about the guidelines for the timeout periods, see 13.7.11(1) Guidelines for the timeout values.

    PATH=/sbin:/bin:/usr/bin:/opt/jp1base/bin:/opt/jp1cons/bin:/opt/jp1imm/bin/imdb
    LANG=C
    MAX_RETRY=24     # 5 * 24 = 120 [seconds]

    Edit the underlined part.

    #1

    The MAX_RETRY value is the number of retries. The retry interval is set to 5 seconds. The MAX_RETRY value is calculated with the following formula:

    [timeout-in-seconds] / 5 = [value-for-MAX_RETRY]

    Example: The timeout in seconds is 180 seconds:

    180 / 5 = 36

    MAX_RETRY = 36

(b) Configure jco_stop

To configure the jco_stop command, follow these steps:

  1. Copy jco_stop.model to a file with a given name.

    cd /etc/opt/jp1cons
    cp -p jco_stop.model given-name
  2. Open and edit the copied script in a text editor.

    To change the timeout value, change the value of MAX_RETRY#1. For details about the guidelines for the timeout periods, see 13.7.11(1) Guidelines for the timeout values.

    PATH=/sbin:/bin:/usr/bin:/opt/jp1base/bin:/opt/jp1cons/bin:/opt/jp1imm/bin/imdb
    LANG=C
    MAX_RETRY=24     # 5 * 24 = 120 [seconds]

    Edit the underlined part.

    #1

    The MAX_RETRY value is the number of retries. The retry interval is set to 5 seconds. The MAX_RETRY value is calculated with the following formula:

    [timeout-in-seconds] / 5 = [value-for-MAX_RETRY]

    Example: The timeout in seconds is 240 seconds:

    240 / 5 = 48

    MAX_RETRY = 48

(4) Setting the timeout values for automatic startup and automatic stop

If JP1/IM - Manager is automatically started and stopped through the automated startup service (systemd), specify the timeout values in the following steps:

  1. Open jp1_cons.service in a text editor.

  2. Edit the following file:

    /usr/lib/systemd/system/jp1_cons.service
  3. In the [Service] section, specify the timeout values.

    Specify TimeoutStartSec=x as the timeout value for startup and TimeoutStopSec=y as the one for stop.

    • When TimeoutStartSec and TimeoutStopSec are not specified, the default timeout period of systemd, 90 seconds, is used.

    • For details about the guidelines for the timeout periods, see 13.7.11(1) Guidelines for the timeout values.

    • For details about how to calculate the operation time for startup or stop from log information at startup or stop of JP1/IM - Manager, see 13.7.11(2) How to measure the startup or stop time from log information.

    • When you change the timeout values of systemd, you need to review the timeout values of the jco_start and jco_stop commands, together with the timeout values of systemd.

  4. Execute the following command to apply the change in the automated startup service (systemd):

    #systemctl daemon-reload