Hi,
We are using BackupExec 12.5 with latest patches (and latest patches applied to the remote agents) and all of the servers we are backing up are virtual servers run from Microsoft's Hyper V server. The server running BackupExec is Windows 2000, one server being backed up via remote agent is Server 2008 and all others are Server 2003.
The problem we are experiencing is that every few weeks (sometimes less) the backup hangs and we find that the remote agent on one particular server still says "snapshot in progress" which is presumably the point where it is taking the system state snapshot. The only way we can get the backup to unfreeze is to restart the Remote Agent service on the server where it has hung. Originally this was only happening on one particular server so we assumed it was an issue with that one server, so we took System State out of the selections list on that server and that seemed to stop the problem from happening. However, today the backup hung again and it was a different server that the agent had hung on... so it seems this is just an issue with the remote agent software, not the servers.
There are no errors in the event logs on any of the servers and the backupexec job log usually just abruptly stops with no error but last night we did actually get an error and it was:
AOFO: Initialization failure on: "D:". Advanced Open File Option used: Microsoft Volume Shadow Copy Service (VSS).
V-79-10000-11233 - VSS Snapshot error. A Microsoft Volume Shadow Copy Service (VSS) snapshot is already running on the target computer. Only one snapshot can run at a time. Try running the job later.
I've tried investigating the issue whilst the the agent is in the hung state and found that the agent process is not using any CPU, memory usage is not changing at all, it still has a couple of TCP connections established to the BackupExec server but is not actually sending anything down them. It is not trying to read from any registry keys or read any files, so its not as if it has just got stuck in a loop trying to read something. The only thing I did spot was that the I/O for the remote agent process is more up and down compared to the agent on other servers, but that might just be general I/O overhead from having the TCP/IP connection open, I dont know.
Anyone got any ideas of how we can fix this?
Thanks
Chris
