4.10 Notes regarding operation

• For notes on executing the jpctool service list command for the Agent Collector service or Agent Store service, see the chapter that describes commands in the manual JP1/Performance Management Reference.

• For notes on executing the jpcspm start command or the jpcspm stop command for PFM - Agent for Platform, see the chapter that describes commands in the manual JP1/Performance Management Reference.

• Hitachi Tuning Manager cannot monitor Performance Management.

• Due to the effects of a virus detection program, file access permission for the files and folders used by Performance Management might be locked by the exclusion control function. As a result, the following symptoms might occur:

  • Performance Management cannot be started.

  • Performance information cannot be collected, or collection is delayed.

  • A Performance Management command terminates abnormally.

  • Logs cannot be output, or a problem cannot be investigated.

  If you want to check viruses while Performance Management is running, deselect the files (including the files in the new storage location if you changed the storage location of the Store database) under the installation folder of Performance Management.

  If you want to check viruses when Performance Management stops and then to restart Performance Management, confirm that the virus check for the files under the installation folder of Performance Management (including the files in the new storage location if you changed the storage location of the Store database) is complete.

• If automatic startup of services is set in a version earlier than 06-70-/I or 07-00-/G, the following problem may occur:

  If PFM service terminates abnormally, for example due to a power failure, rebuilding the Store database index file when starting the Agent Store service takes a long time At that time, Agent Collector service may fail to start because it cannot communicate with the Agent Store service.

  In order to prevent this problem, it is recommended to install PFM - Agent Product of version 06-70-/I or 07-00-/G or later, because the contents of the automatic startup script model file has changed. Please make a backup copy of the automatic startup script file which is currently used, and change its contents by using the new model file. For details on how to do this, see the chapter that explains how to enable or disable the automatic service start feature for monitoring managers and monitoring agents (UNIX) in the JP1/Performance Management User's Guide.

  In addition, after resetting the automatic startup script file, service startup operations will change as follows:

  • Because the PFM service startup processes are executed in the background, these services might still be starting up even after the OS startup processing has completed.

  • Since each Performance Management service is synchronized before starting up, startup time might increase.

  • Before reset, the following message is no longer output to the console. Please see the common message log to confirm that the PFM service have started.

    KAVE06007-I The service will now start. (service=service-name, lhost=logical-host-name, inst=instance-name)

• In HP-UX, even when the Node and Host Name Expansion function is enabled, the length of host names that can be used in Performance Management is limited to 8 bytes from the start of the string. If you need to run Performance Management on a host whose name is more than 8 bytes long, see the chapter that presents notes regarding operation in the JP1/Performance Management User's Guide.

• In HP-UX and Solaris, the automatic start function is valid when the PFM service is started on an OS in Runlevel 2. The automatic stop function is valid when the PFM service is started on an OS in Runlevel 1.

  In Linux, the automatic start function is valid when the PFM service is started on an OS in Runlevel 3 or 5. The automatic stop function is valid when the PFM service is started on an OS in Runlevel 0 or 6.

  When the PFM service is started in other Runlevels, neither the automatic start nor the automatic stop function operates.

• In HP-UX, the operation of PFM - Agent for Platform in high reliability mode is not supported.

• When PD_FSL and PD_FSR are collected, system calls are issued at collection-time intervals and information is collected. Because file systems are accessed in order to collect information, if you attempt to use the umount command to unmount the file system while information is being collected, the execution of the umount command might fail. In such instances, execute the umount command again at a time when information is not being collected.

• In Linux, PFM - Agent for Platform does not support environments where the number of CPUs changes dynamically. If the number of CPUs changes, the Agent Collector service must be restarted.

• When the NFS File System property is not set to No, operate PFM - Agent for Platform in the status in which information for the mounted remote file system can be referenced (status in which df command can be executed normally). If PD_FSL, PD_FSR records are collected when the mounted remote file system has not responded, the Agent Collector service hangs up, making it impossible to continue collection of all performance data.

  When data collection stopped while PD_FSL or PD_FSR records were being collected:

  1. Use the kill command to stop the jpcagtu process.

     # kill -TERM (or KILL) <jpcagtu-process-ID>

  2. Recover the status in which the remote file system is normally mounted. For example, restart the NFS daemon.

  3. Use the jpcspm start command to start the Agent Collector service.

     # /opt/jp1pc/tools/jpcspm start -key agtu

• In AIX, in version PFM - Agent for Platform 08-00 and later, since the information of non-operational CPU can be allocated to the system, and is added as an instance, hence there can be cases where the instance count of PI_CPUP record would have increased, when compared with versions prior to PFM - Agent for Platform 08-00.

• If the Emulex Fibre Channel card virtualization function and N_Port ID Virtualization (NPIV) function that are provided in HP-UX are used to dynamically change a disk device, the following problems sometimes occurs in the Physical Disk Overview (PI_PDSK) record:

  • Performance data for a dynamically added disk device is collected only when I/O operations are being performed and not collected during other periods.

  • When delta values are to be indicated in real-time reports, an invalid value is displayed in the performance data that was collected during the first data collection operation for a dynamically added disk device.

  If such problems occur, consider operating the system by using the following methods:

  • To evaluate an alarm, remove any dynamically added disk devices from the monitoring target.

  • Make sure Indicate delta value is not selected in the window for configuring the report display settings. Alternatively, to check the delta value of a dynamically added disk device, refer to the performance data collected during the second or a later data collection operation.

To Page Top