PRIMECLUSTER Global Disk Services Configuration and AdministrationGuide 4.7

6.12.2 Hang-up Monitoring Function

Because the GDS I/O Monitor Option assures the I/O response time, it is equipped with hang-up monitoring functions to monitor whether or not the GDS daemon or driver is hung-up.

The hang-up monitoring function monitors whether or not the sdxservd daemon and the sfdsk driver of the current node are hung-up, and if a hang-up is detected it panics the current node.

The hang-up monitoring function checks the status of the daemon and driver at a 154 seconds monitoring interval, if there is no response within the 77 second conclusion time it will be determined to be hung-up.

If a value other than the default value is set for the I/O response time, change the monitoring interval and conclusion time parameter of the hang-up monitoring function as follows:

Item	Parameter Name	Explanation
Monitoring Interval	SDX_HCHK_INTERVAL	The time interval (seconds) after monitoring has finished before the next monitoring start. Set to double the time of the conclusion time. Default Value: 154 Possible setting values: 12 or more, 2592000 or less, and a value greater than that of the conclusion time (seconds).
Conclusion Time	SDX_HCHK_CRITERION	Time (seconds) after monitoring has been started before it is judged to be hung-up. Set to the same value as the I/O response time. If the value of the I/O response time differs depending on the class, set the shortest time. Default value: 77 Possible setting values: 1 or more, 3600 or less, and a value less than the monitoring interval (seconds).

How to change parameters

1. Change the setting parameters in the GDS settings file /etc/opt/FJSVsdx/sdx.cf, or add parameter settings.

   Example) If setting the monitoring time interval to 156 seconds and the conclusion time to 78 seconds

   SDX_HCHK_INTERVAL=156

   SDX_HCHK_CRITERION=78

2. Restart the system. Or execute the following command.

   # /usr/bin/systemctl restart fjsvsdxhealth.service

3. After restarting the system or service with step 2, make sure that the service is operating.

   # /usr/bin/systemctl status fjsvsdxhealth.service
   * fjsvsdxhealth.service - PRIMECLUSTER Global Disk Services hang up monitor service
      ...
      Active: active (running) ...
   ...

Note

No monitoring is done in single user mode.