Running a three node cluster on HPUX, 11.31. We have the vxconfigd daemon running, but something i see is that the daemon on one server uses about 20-50% CPU util, on another server about the same, and then on the third server of the cluster very little at any given moment. Anyone have any ideas as to why this might be the case? it doesn't really bother me, but the DBA's get all concerned about it. Any info would be appreciated.
Well I don't believe we can point out any straight reason without appropriate investigation. There could be n number of possibilities. I would though suggest to look into following
1. Is the utilization going up in specific trend or its once off ? what process consumes highest CPU at the point ?
2. While you are noticing high utilization, what is trend of application running ? Is it expected to write/read heavy IOs on the volumes ?
3. Whether utilization is increasing at any specific time ? If yes, is there any planned snapshots / backups etc happening ?
4. Do you have VVR replication as well ? VVR code is integrated in vxconfigd ... Any replication going on ?
5. Do you happen to notice any errors/messages in logs while utilization is high ?
6. Any difference in storage lun presentations between 3 nodes ?
7. All the products including OS, Application, SF are at right/updated level of patches ?
There could be many more thoughts around ...
Quite possible that vxconfigd is reacting to what is happening to applications/infrastructure .. would suggest to pin down the process consuming CPU & trend of the CPU utilization to get to exact root cause ..