I've recently been workign with vcb and netbackup to backup my vm's, but I've had some problems on some of my vm workstations. If the snapshot failed for any reason, then it would get stuck on the vm, without deleting. It also caused my domain controller to completely lose itself. I had to reboot it for it to start working again. Anyway, I started working with just the vcbmounter and it seems to have stopped an oracle database on one of my servers, and messed up some tables on one of my sql servers.
I am basically running these vcb backups with default settings, and not freeze scripts. I'm assuming that the problems I'm seeing are due to the quiesce step. I was just wondering if it's not recommended for oracle, sql, domain controllers.
Image based backup is not a good idea for Domain Controllers. Google for USN rollback. I've also read that VCB backups are crash consistent backups which is not a good idea for databases.
as long as your ESX is 3.5 U1 and above, the VMtools that correspond with the above, and the OS is windows 2003 and the application running inside can utilize VSS, you can make an application consistent backup..so in your example, i would make sure my vmtools have the vss components installed and re-run the backup..it will work fine this tme..
where Information Lives
Thanks for the info Itzikr. I have everything you listed, except I don't know if the application supports vss. I did some testing last night on one of my 2003 servers running an Oracle DB and found that after 3 vcb snapshots the oracle database would shut down. if I ran the vcb snapshot with the -Q 0 (no quiesce) option then it would not shut down the database.
This leads me to believe that the quiesce is what's causing my servers to have problems, and I'm wondering if there are any issues with me running all of my snapshots with the -Q 0. Does it mean I just won't get files that are currently open? Basically I have about 60 vm's and I don't want to go through each and every one of them to find out if they can survive a snapshot or not, and I also don't want to just run the snapshots and break my vm's...
I have our DBA's write quiesce/unquiesce routines to execute before and after Oracle backups. VCB has hosed this server before (I guess transactions are faster than snapshot merges.) We also have DB2 and SQL databases, but after the Oracle incident I don't use VCB on them (but I'm sure you could.)
Oracle will eventually have a problem if you don't gracefully stop and start it accordingly.