Ran across this issue. Have an open ticket. Not sure if it's a bug in DP or something on my end. Sharing in the event it can save someone else a headache.
On a Monday 2 weeks ago, I updated my DP 8.14 environment to 9.0 and then patched to 9.06. Everything went fine and ran without issue for 2 days. I have a Monday morning scheduled BRICK (brick --full) run via cron. I decided to run the BRICK report Tuesday afternoon on the way out the door. I came back in Wednesday morning and found that I had several jobs that had stopped on on initializing medium to a file library device and they were In Progress with no activity and jobs that got queued after the devices were used up timed out without issue.
I tried to abort the jobs from Monitor tab of the GUI client (Windows), which resulted in the spinning blue mouse pointer and a "(Not Responding") on the window. Attempting an omniabort -session on the commandline just hung too. I ended up using the omnisv to stop all services and start them back up. The GUI would report the jobs failed but a "ps -ef | grep omni" would show bsm jobs still running. I killed them off and attempted to perform backups again and got intermittent results getting some jobs to go through and then it'd lock up again.
Creating Debug sessions would generate files on the CM and MA but not the DA. I ran Wireshark on the DA and MA and traffic was apparent from CM to DA and CM to MA but nothing between the DA to MA. I attempted to gather debugs and the debug collection would just hang and not show any activity and require me to manually gather them.
I initially didn't put 2 and 2 together and point the finger at the BRICK reports. I got it up and running and did try to run a omnicheck -dns - full after HPE Support suggested there was a DNS issue and noticed the bsm process hang again. Further omnidlc debug collections would hang after that too.
The eventual realization that the omnicheck -full -dns, which BRICK runs in it's report gathering, was what was causing backup jobs to stop functioning has led to me removing my cron job. I plan to use a config file for BRICK to skip the dns check but still gather some reports.
Regarding DNS check - it worked fine on 8.14. I was notified by HPE support that one of my debugs gathered had a bad reverse DNS on a client, however, when DP was running it would run the backup job on that client without issue.
During the past 2 weeks I uninstalled the Patch, reinstall 9 and then repatch it without success. Reboot the server several times. Try multiple omnirc options, change global options, send in multiple logs to HPE.
Resolution when broken to get it going again:
killall bsm
killall omnidlc
omnisv -stop
omnisv -start
run a debug delete on all clients - any of them that it hangs on need to have the INET service restarted, seemed to be only Windows machines - I think these bad INETs resulted in jobs that kicked off when the CM was hosed and it broke the DA clients.
Here's the big one: Don't run omnicheck -dns
Thoughts
I'm not 100% sure, but it seems to me that the busted INET services on the DA's would cause a restart/scheduled job on that client to fail again and continue the CM in it's broken loop for future jobs on all other clients. Killing the bsm and omnidlc processes that are hung and restarting all the INET services that don't respond to the Debug Delete seems to bring all within line, at least the 1.5 days of green check marks in Sessions seems to concure with it.
Technical Information:
BRICK v 2.01 from www.data-protector.org
Cell Manager
Hardware: HP DL320e Server
Software:: centos-release-6-7.el6.centos.12.3.x86_64
/omnicheck -patches
Patch level Patch description
===========================================
DPUX_00175/DPLNX_00464(BDL906) Cell Manager Component
DPUX_00175/DPLNX_00464(BDL906) Cell Server Technology Stack
DPUX_00175/DPLNX_00464(BDL906) Application Server Technology Stack
DPUX_00175/DPLNX_00464(BDL906) Web Services
DPUX_00175/DPLNX_00464(BDL906) Java Runtime Environment Technology Stack
DPUX_00175/DPLNX_00464(BDL906) Job Control Engine Service Dispatcher
DPUX_00175/DPLNX_00464(BDL906) Job Control Engine Service Registry
Number of patches found: 7.
2016-05-27 3:11pm EST - Update
I ran BRICK with as "brick --full -n --path=/var/www/hostname/public_html" with the "-n" meaning no dns check and it completed with out issue and didn't break my backups and I've run 3 backups since it completed and all looks good.