VOMS Continously Being Restarted By Monitoring

Start Time	September 30th , 22:00	voms being restarted by monitoring
Change	October 4th, 14:00	monitoring time out increased, service no longer restarting
Current Status	Service is stable but requests are much slower

Initial Observation

All three active voms servers have been suffering such that the monitoring detects an error with voms and then restarts the affected voms instances.

Easy Checks.

1. Database group contacted - nothing different or interesting at their end. There is nothing on the service to suggest this.
   • Have replied that service running normally.
2. vomrs stopped for a period, this would have stopped registration process but is an easy item to eliminate.
3. No packages have been updated in this time frame.
4. No CDB changes have been made at this time.
5. No operators or other admins have logged in at this time.

VomsLoad Metric

The VomsLoad metric runs voms-ping. This essentially tries to generate a proxy from each of the VOs. From 22:00 on the 30th this started to take much longer to run.

From the afternoon of the 4th of October the monitoring was altered to allow this longer requests to complete.

Possibilities to Check.

• Andrew Elwell suggested an extra 2 seconds smells of DNS timeout - to be checked.

-- SteveTraylen - 2009-10-01