vxdbd high cpu usage @ hpux
There is evidence vxdbd daemon is consuming too much cpu and following its behaviour has many EPIPE errors on writes.
See first column:
$ grep " write" ./pid.12331 | awk '{print $4,$5, $7, $9, $16,$17}' | sort | uniq -c | sort -rn
45355 pid=12331 ktid=20116 write err=32 A0=0 A1=6
45219 pid=12331 ktid=20117 write err=32 A0=0 A1=10
43862 pid=12331 ktid=20118 write err=32 A0=0 A1=11
40211 pid=12331 ktid=20121 write err=32 A0=0 A1=13
39893 pid=12331 ktid=20120 write err=32 A0=0 A1=12
We think restarting the service could eliminate such condition. Of course the above as a workaround. There is a patch which has some fixes for it , i.e.
PHCO_41073;
Before trying a restart of the service i.e. as per manual documentation:
To start the vxdbd daemon
Use the vxdbdctrl start command:
/opt/VRTSdbcom/bin/vxdbdctrl start
To stop the vxdbd daemon
As root, use the vxdbdctrl stop command:
/opt/VRTSdbcom/bin/vxdbdctrl stop
Our question is:
Q. Is it safe to restart vxdbd daemon with no colateral impact on production services due the nature or extension of vxdbd's functionality?
Best Regards.
Hi Alizarra,
Restarting vxdbd should be safe under normal conditions. This daemon is responsible for communication
between SF-ORA products. Stopping this daemon should not cause any I/O issues to database or filesystems.
Hope this helps.
Srini