Hi,
I have NetBackup 6.0MP2 installed on the master server which also happens to be the media server (an RHEL AS 4 server). The client is a Red Hat Enterprise Linux system, and the requirement is to backup a directory containing several sub-directories and files within them. I have created a policy to this effect.
When I choose a manual backup of this policy, and have a test set of around 10GB of files under that directory on the client, the nbproxy process takes a long time (as much as 4 hours) to run, and no activity is shown in the Activity Monitor during this time. After this, the actual backup starts and I am able to see the little green running men in the Activity Monitor.
While the time taken to backup is nothing to write home about, I am curious why nbproxy shows a 98% CPU usage (from the top command). I turned on the nbproxy legacy log, and see this message repeat continuously:
14:54:29.442 <2> db_OpenImgWithLockByPath: trying to lock : db_Imglock(83463c8) offset 0 in /usr/openv/netbackup/db/images/client.example.com/1146000000/tmp/ATP-gp-media_1146009145_FULL
14:54:29.442 <2> db_OpenImgWithLockByPath: locked : db_Imglock(83463c8): offset 0 in /usr/openv/netbackup/db/images/client.example.com/1146000000/tmp/ATP-gp-media_1146009145_FULL
On the Veritas support site, I found this document:
http://seer.support.veritas.com/docs/280757.htm
which speaks of a deadlock in the bpdbm. Is there any relation between this and nbproxy, causing it to show high CPU usage?
Thanks in advance,
Sandeep
--
http://sandeep.weblogs.us/