Top
Systemwalker Operation Manager  Cluster Setup Guide for UNIX
FUJITSU Software

2.2.3 Canceling the settings for starting and stopping daemons automatically

After installing Systemwalker Operation Manager, cancel the settings that start and stop the daemons automatically.

Stopping daemons

Stop each Systemwalker Operation Manager daemon on all the nodes (active and standby). The following example shows the stop command:

# /opt/systemwalker/bin/poperationmgr -s

For environments that include both Systemwalker Operation Manager and Systemwalker Centric Manager, be sure to specify the "-s" option with the poperationmgr command.

Canceling the settings to start and stop daemons automatically

On each node (active and standby), cancel the automatic startup settings for the following Systemwalker Operation Manager daemons:

Canceling the settings for starting daemons automatically

To cancel the settings for starting daemons automatically, use the following procedure to edit the daemon batch start customization file:

For PRIMECLUSTER (for Solaris or Linux version), Oracle Solaris Cluster

  1. Open the following daemon batch start customization file with a text editor such as vi:

    /etc/opt/FJSVftlo/daemon/custom/rc3.ini
  2. Comment out the lines shown in the following examples (from [Before] to [After]):

    [Before]
    @   Function name: Job Execution Control
    (data omitted...)
    DAEMONxx(Lo)="/etc/opt/FJSVMJS/etc/rc3.d/S99MJS"
    
    Note: "xx" is a number.
    [After]
    @   Function name: Job Execution Control
    (data omitted...)
    #DAEMONxx(Lo)="/etc/opt/FJSVMJS/etc/rc3.d/S99MJS"
    Note: "xx" is a number.
    [Before]
    @   Function name: Jobscheduler
    (data omitted...)
    DAEMONxx(Lo)="/opt/FJSVJOBSC/etc/rc3.d/start_jobs"
    
    Note: "xx" is a number.
    [After]
    @   Function name: Jobscheduler
    (data omitted...)
    #DAEMONxx(Lo)="/opt/FJSVJOBSC/etc/rc3.d/start_jobs"
    Note: "xx" is a number.
  3. When a 1:1 active/standby configuration (with subsystems and partial cluster operation) is to be used, edit the settings so that the subsystems that are not registered with a cluster system start automatically.

    The following example shows an environment containing two subsystems, Subsystem 0 and Subsystem 1. This example illustrates the procedure used to register Subsystem 1 with a cluster system while leaving Subsystem 0 not registered.

    [Before]
    @   Function name: Job Execution Control
    (data omitted...)
    #DAEMONxx(Lo)="/etc/opt/FJSVMJS/etc/rc3.d/S99MJS"
    
    Note: "xx" is a number.
    [After]
    @   Function name: Job Execution Control
    (data omitted...)
    #DAEMONxx(Lo)="/etc/opt/FJSVMJS/etc/rc3.d/S99MJS"
    DDAEMONxx0(Lo)="/etc/opt/FJSVMJS/etc/rc3.d/S99MJS -sys 0"
    Note: "xx" is a number.
    [Before]
    @   Function name: Jobscheduler
    (data omitted...)
    #DAEMONxx(Lo)="/opt/FJSVJOBSC/etc/rc3.d/start_jobs"
    
    Note: "xx" is a number.
    [After]
    @   Function name: Jobscheduler
    (data omitted...)
    #DAEMONxx(Lo)="/opt/FJSVJOBSC/etc/rc3.d/start_jobs"
    DAEMONxx0(Lo)="/opt/FJSVJOBSC/etc/rc3.d/S99JOBSCH -sys 0"
    Note: "xx" is a number.

    Note the following points with regard to step 3:

    • When naming "DAEMONxx(Lo)" variables, ensure that the subsystem number that will be subject to automatic startup is appended to the end of the numeric part of the name. For example, if the variable name is "DAEMON05(Lo)" and the subsystem is "0", add "0" to the end so that the variable name becomes "DAEMON050(Lo)".

    • Specify the number of the subsystem that will be subject to automatic startup after the -sys option.

    • If there is more than one subsystem that will not be registered with a cluster system, add a definition for all of the subsystems concerned. (The specification can extend over multiple lines.)

For MC/ServiceGuard

  1. Open the following daemon batch start customization file with a text editor such as vi:

    /etc/opt/FJSVftlo/daemon/custom/rc3.ini
  2. Comment out the lines shown in the following examples (from [Before] to [After]):

    [Before]
    @   Function name: Job Execution Control
    (data omitted...)
    DAEMONxx(Lo)="/opt/FHPMJS/usr/lib/mjes/S99MJS"
    
    Note: "xx" is a number.
    [After]
    @   Function name: Job Execution Control
    (data omitted...)
    #DAEMON0xx(Lo)="/opt/FHPMJS/usr/lib/mjes/S99MJS"

    Note: "xx" is a number.
    [Before]
    @   Function name: Jobscheduler
    (data omitted...)
    DAEMON0xx(Lo)="/opt/FHPJOBSCH/bin/start_jobs"
    
    Note: "xx" is a number.
    [After]
    @   Function name: Jobscheduler
    (data omitted...)
    #DAEMONxx(Lo)="/opt/FHPJOBSCH/bin/start_jobs"
    Note: "xx" is a number.
  3. In the case of the 1:1 active/standby configuration (with subsystems and partial cluster operation), edit the settings so that subsystems that are not registered in clusters will be started automatically.

    The following example illustrates the procedure used to register Subsystem 1 but not Subsystem 0 in a cluster, in an environment in which Subsystems 0 and 1 have been created.

    [Before]
    @   Function name: Job Execution Control
    (data omitted...)
    #DAEMONxx(Lo)="/opt/FHPMJS/usr/lib/mjes/S99MJS"
    
    Note: "xx" is a number.
    [After]
    @   Function name: Job Execution Control
    (data omitted...)
    #DAEMONxx(Lo)="/opt/FHPMJS/usr/lib/mjes/S99MJS"
    DAEMONxx0(Lo)="/opt/FHPMJS/usr/lib/mjes/S99MJS -sys 0"
    Note: "xx" is a number.

    [Before]
    @   Function name: Jobscheduler
    (data omitted...)
    #DAEMONxx(Lo)="/opt/FHPJOBSCH/bin/start_jobs"
    
    Note: "xx" is a number.
    [After]
    @   Function name: Jobscheduler
    (data omitted...)
    #DAEMONxx(Lo)="/opt/FHPJOBSCH/bin/start_jobs"
    DAEMONxx0(Lo)="/opt/FHPJOBSCH/bin/S99JOBSCH -sys 0"
    Note: "xx" is a number.

    Note the following points with regard to step 3:

    • When naming "DAEMONxx(Lo)" variables, ensure that the subsystem number that will be subject to automatic startup is appended to the end of the numeric part of the name. For example, if the variable name is "DAEMON05(Lo)" and the subsystem is "0", add "0" to the end so that the variable name becomes "DAEMON050(Lo)".

    • Specify the automatic start target subsystem number as the -sys option.

    • If there are multiple subsystems that will not be registered in the cluster, add a definition for all the subsystems that will not be registered. (This can be defined on multiple lines.)

For HACMP

  1. Open the following daemon batch start customization file with a text editor such as vi:

    /etc/opt/FJSVftlo/daemon/custom/rc3.ini
  2. Comment out the lines shown in the following examples (from [Before] to [After]):

    [Before]
    @   Function name: Job Execution Control
    (data omitted...)
    DAEMONxx(Lo)="/opt/FAIXMJS/usr/lib/mjes/S99MJS"
    
    Note: "xx" is a number.
    [After]
    @   Function name: Job Execution Control
    (data omitted...)
    #DAEMONxx(Lo)="/opt/FAIXMJS/usr/lib/mjes/S99MJS"
    Note: "xx" is a number.

    [Before]
    @   Function name: Jobscheduler
    (data omitted...)
    DAEMONxx(Lo)="/usr/FAIXJOBSC/bin/start_jobs"
    
    Note: "xx" is a number.
    [After]
    @   Function name: Jobscheduler
    (data omitted...)
    #DAEMONxx(Lo)="/usr/FAIXJOBSC/bin/start_jobs"
    Note: "xx" is a number.
  3. In the case of the 1:1 active/standby configuration (with subsystems and partial cluster operation), edit the settings so that subsystems that are not registered in clusters will be started automatically.

    The following example illustrates the procedure used to register Subsystem 1 but not Subsystem 0 in a cluster, in an environment in which Subsystems 0 and 1 have been created.

    [Before]
    @   Function name: Job Execution Control
    (data omitted...)
    #DAEMONxx(Lo)="/opt/FAIXMJS/usr/lib/mjes/S99MJS"
    
    Note: "xx" is a number.
    [After]
    @   Function name: Job Execution Control
    (data omitted...)
    #DAEMONxx(Lo)="/opt/FAIXMJS/usr/lib/mjes/S99MJS"
    DAEMONxx0(Lo)="/opt/FAIXMJS/usr/lib/mjes/S99MJS -sys 0"
    Note: "xx" is a number.

    [Before]
    @   Function name: Jobscheduler
    (data omitted...)
    #DAEMONxx(Lo)="/usr/FAIXJOBSC/bin/start_jobs"
    
    Note: "xx" is a number.
    [After]
    @   Function name: Jobscheduler
    (data omitted...)
    #DAEMONxx(Lo)="/usr/FAIXJOBSC/bin/start_jobs"
    DAEMONxx0(Lo)="/usr/FAIXJOBSC/bin/S99JOBSCH -sys 0"
    Note: "xx" is a number.

    Note the following points with regard to step 3:

    • When naming "DAEMONxx(Lo)" variables, ensure that the subsystem number that will be subject to automatic startup is appended to the end of the numeric part of the name. For example, if the variable name is "DAEMON05(Lo)" and the subsystem is "0", add "0" to the end so that the variable name becomes "DAEMON050(Lo)".

    • Specify the automatic start target subsystem number as the -sys option.

    • If there are multiple subsystems that will not be registered in the cluster, add a definition for all the subsystems that will not be registered. (This can be defined on multiple lines.)

Canceling the settings for stopping daemons automatically

To cancel the settings for stopping daemons automatically, use the following procedure to edit the daemon batch stop customization file:

For PRIMECLUSTER (for Solaris or Linux version), Sun Cluster

  1. Open the following daemon batch stop file with a text editor such as vi:

    /etc/opt/FJSVftlo/daemon/ini/rc0.ini
  2. Comment out the lines shown in the following examples (from [Before] to [After]):

    [Before]
    DAEMONxx(Lo)="/opt/FJSVJOBSC/etc/rc0.d/end_jobs"
    DAEMONxx(Lo)="/etc/opt/FJSVMJS/etc/rc0.d/K00MJS"
    Note: "xx" is a number.
    [After]
    #DAEMONxx(Lo)="/opt/FJSVJOBSC/etc/rc0.d/end_jobs"
    #DAEMONxx(Lo)="/etc/opt/FJSVMJS/etc/rc0.d/K00MJS"
    Note: "xx" is a number.
  3. If a 1:1 active/standby configuration (with subsystems and partial cluster operation) is to be used, edit the settings so that subsystems that will not be registered with a cluster system are shut down automatically.

    The following example shows an environment containing two subsystems, Subsystem 0 and Subsystem 1. This example illustrates the procedure used to register Subsystem 1 with a cluster system while leaving Subsystem 0 not registered.

    [Before]
    #DAEMONxx(Lo)="/opt/FJSVJOBSC/etc/rc0.d/end_jobs"
    #DAEMONxx(Lo)="/etc/opt/FJSVMJS/etc/rc0.d/K00MJS"
    Note: "xx" is a number.
    [After]
    #DAEMONxx(Lo)="/opt/FJSVJOBSC/etc/rc0.d/end_jobs"
    DAEMONxx0(Lo)="/opt/FJSVJOBSC/etc/rc0.d/K00JOBSCH -sys 0"
    #DAEMONxx(Lo)="/etc/opt/FJSVMJS/etc/rc0.d/K00MJS"
    DAEMONxx0(Lo)="/etc/opt/FJSVMJS/etc/rc0.d/K00MJS -sys 0"
    Note: "xx" is a number.

    Note the following points with regard to step 3:

    • When naming "DAEMONxx(Lo)" variables, ensure that the subsystem number that will be subject to automatic shutdown is appended to the end of the numeric part of the name. For example, if the variable name is "DAEMON02(Lo)" and the subsystem is "0", add "0" to the end so that the variable name becomes "DAEMON020(Lo)".

    • Specify the number of the subsystem that will be subject to automatic shutdown after the -sys option.

    • If there is more than one subsystem that will not be registered with a cluster system, add a definition for all of the subsystems concerned. (The specification can extend over multiple lines.)

For MC/ServiceGuard

  1. Open the following daemon batch stop file with a text editor such as vi:

    /etc/opt/FJSVftlo/daemon/ini/rc0.ini
  2. Comment out the lines shown in the following examples (from [Before] to [After]):

    [Before]
    DAEMONxx(Lo)="/opt/FHPJOBSCH/bin/end_jobs"
    DAEMONxx(Lo)="/opt/FHPMJS/usr/lib/mjes/K00MJS"
    Note: "xx" is a number.
    [After]
    #DAEMONxx(Lo)="/opt/FHPJOBSCH/bin/end_jobs"
    #DAEMONxx(Lo)="/opt/FHPMJS/usr/lib/mjes/K00MJS"
    Note: "xx" is a number.
  3. In the case of the 1:1 active/standby configuration (with subsystems and partial cluster operation), edit the settings so that subsystems that are not registered in clusters will be stopped automatically.

    The following example illustrates the procedure used to register Subsystem 1 but not Subsystem 0 in a cluster, in an environment in which Subsystems 0 and 1 have been created.

    [Before]
    #DAEMONxx(Lo)="/opt/FHPJOBSCH/bin/end_jobs"
    #DAEMONxx(Lo)="/opt/FHPMJS/usr/lib/mjes/K00MJS"
    Note: "xx" is a number.
    [After]
    #DAEMONxx(Lo)="/opt/FHPJOBSCH/bin/end_jobs"
    DAEMONxx0(Lo)="/opt/FHPJOBSCH/bin/K00JOBSCH -sys 0"
    #DAEMONxx(Lo)="/opt/FHPMJS/usr/lib/mjes/K00MJS"
    DAEMONxx0(Lo)="/opt/FHPMJS/usr/lib/mjes/K00MJS -sys 0"
    Note: "xx" is a number.

    Note the following points with regard to step 3:

    • When naming "DAEMONxx(Lo)" variables, ensure that the subsystem number that will be subject to automatic shutdown is appended to the end of the numeric part of the name. For example, if the variable name is "DAEMON02(Lo)" and the subsystem is "0", add "0" to the end so that the variable name becomes "DAEMON020(Lo)".

    • Specify the automatic stop target subsystem number as the -sys option.

    • If there are multiple subsystems that will not be registered in the cluster, add a definition for all the subsystems that will not be registered. (This can be defined on multiple lines.)

For HACMP

  1. Open the following daemon batch stop file with a text editor such as vi:

    /etc/opt/FJSVftlo/daemon/ini/rc0.ini
  2. Comment out the lines shown in the following examples (from [Before] to [After]):

    [Before]
    DAEMONxx(Lo)="/usr/FAIXJOBSC/bin/end_jobs"
    DAEMONxx(Lo)="/opt/FAIXMJS/usr/lib/mjes/K00MJS"
    Note: "xx" is a number.
    [After]
    #DAEMONxx(Lo)="/usr/FAIXJOBSC/bin/end_jobs"
    #DAEMONxx(Lo)="/opt/FAIXMJS/usr/lib/mjes/K00MJS"
    Note: "xx" is a number.
  3. In the case of the 1:1 active/standby configuration (with subsystems and partial cluster operation), edit the settings so that subsystems that are not registered in clusters will be stopped automatically.

    The following example illustrates the procedure used to register Subsystem 1 but not Subsystem 0 in a cluster, in an environment in which Subsystems 0 and 1 have been created.

    [Before]
    #DAEMONxx(Lo)="/usr/FAIXJOBSC/bin/end_jobs"
    #DAEMONxx(Lo)="/opt/FAIXMJS/usr/lib/mjes/K00MJS"
    Note: "xx" is a number.
    [After]
    #DAEMONxx(Lo)="/usr/FAIXJOBSC/bin/end_jobs"
    DAEMONxx0(Lo)="/usr/FAIXJOBSC/bin/K00JOBSCH -sys 0"
    #DAEMONxx(Lo)="/opt/FAIXMJS/usr/lib/mjes/K00MJS"
    DAEMONxx0(Lo)="/opt/FAIXMJS/usr/lib/mjes/K00MJS -sys 0"
    Note: "xx" is a number.

    Note the following points with regard to step 3:

    • When naming "DAEMONxx(Lo)" variables, ensure that the subsystem number that will be subject to automatic shutdown is appended to the end of the numeric part of the name. For example, if the variable name is "DAEMON02(Lo)" and the subsystem is "0", add "0" to the end so that the variable name becomes "DAEMON020(Lo)".

    • Specify the automatic stop target subsystem number as the -sys option.

    • If there are multiple subsystems that will not be registered in the cluster, add a definition for all the subsystems that will not be registered. (This can be defined on multiple lines.)

Changing the monitored process

On each node (active and standby), remove the Jobscheduler and Job Execution Control from the list of processes monitored by the process monitoring function.

  1. Open the following definition file for monitored processes with a text editor such as vi:

    /etc/opt/FJSVftlo/pmon/mppmon.usr
  2. Edit the lines shown in the following examples (from [Before] to [After]):

    [Before]
    start_jobs=1
    S99MJS=1
    [After]
    start_jobs=0
    S99MJS=0

Refer to "Changing the Processes to be monitored" in the Systemwalker Operation Manager Installation Guide for details on the procedure for changing the monitored processes.

Note

As a result of this procedure, Jobscheduler and Job Execution Control processes for subsystems not registered with a cluster system will also be removed from the list of processes monitored by the process monitoring function. This procedure will also prevent messages from being sent to the event log or syslog when a process error occurs.