In case anyone is interested. We basically fixed the problem by rebooting one of the media servers. It appears that it was having intermittent difficult, or perhaps one way, communication with the master server. The master server was unable to empty it queue (forgotten the name) L so it looked like everything was hung.
We found the problem from two different angles by two different people.
On the media server we saw very slow responses to vmoprcmd and tpconfig (among others). The most interesting thing was that the media server could ping the physical master server but not the virtual (it’s clustered, both on same subnet). However the master server could ping the media server and other media servers had no problem pinging both. After the reboot, the medias server was able to ping both. L
From another angle but in parallel someone ran some analysis tools on the master server and determined the above queue problem. After rebooting the medias server (it had been up for 45 days (Linux)) everything worked find. We went from a success rate of 65% to 80% (on a good day) to 95% the first night of the fix and 99% last night.
This was all because of one errant media server out of 60. L
Regards,
Patrick Whelan
VERITAS Certified NetBackup Support Engineer for UNIX.
VERITAS Certified NetBackup Support Engineer for Windows.
netbackup AT whelan-consulting.co DOT uk
