Snapshot deleted by external source

[sdime808]

I'm using BE 10d with CPS. In CPS I take snapshots 4 times a day with the last one for each day being saved for a week. But every night all my snapshots get deleted and I get and alert "snapshot deleted by an external source". In my event log I get the same warning and sometime just before that I get an error "Logical drive D: fell below threshold of 801.33 MB. The current value is 740.00 MB." Source is the "Director Agent".

My data is kept on E:, my shadow copied are kept on D:. I have about 20 GB of free space on D:. I've tried changing the shadow copy settings and everything I can think of, and the tech support at Veritas hasn't been very helpful. I am new to shadow copies as you can probably tell. Anyone have any ideas?

 

[steveot]

Take a look at thos technote.

Sounds like you need more space. There should be some VSS events in app/system log.

You may need to get Microsoft involved to determine whey VSS is deleting the snapshots.

http://support.veritas.com/docs/281173

 

[sdime808]

The error I get in the event log is

"Logical drive D: fell below threshold of 801.33 MB. The current value is 740.00 MB." Source is the "Director Agent".

It seems like a space issue to me too, but I have about 75% free space left on the drive where the shadow copies are being saved.

I guess I'll give Veritas another call, then Microsoft if needed. Thanks!

 

[sdime808]

Ok, I talked to tech support and they offer no help in this situation. Its a problem with Microsoft Shadow copy and they said I need to contact them. They basically washed their hands of it. The Tech told me that he's had three reported cases of this today, if you were thinking about using CPS, I'd looking into this before making your decision...

 

[steveot]

They are correct. How is it a Symantec problem if you look in the App/System log and the error are coming from VSS?

You need to figure out why the Snapshots are bing deleted. Basically reduce you snapshot retension and see what happens. At the technote states above what error do you see? Disk I/O..Space?

 

[sdime808]

The reason I feel like they should be more help is, they built their product on top of VSS, if this is a common problem, as the Tech mentioned by the number of cases reported, then they should have tested it more before they released it. Or at least help me figure out where to begin.

I don't think its a retension issue, because its not even able to store more than a days worth of snapshots. I ran Perfmon like the Technote said to do, now I'm just trying to figure out what all the numbers mean.

 

[steveot]

Symantec call is a third party provider that calls on VSS. Many other application can be "requestors" and use the writers. What would symantec do now if they had tested it more if the error again is returning from VSS. There is nothing that can be changed since CPS is just passing the data to VSS and telling it to take a snapshot of this data.

I have seen the problem and experienced it. Its all about timing and setting. When you CPS jobs take place, when the snap shots take place, do you have a continious job that is replicating data to a volume and you snapshots are being performed at the sametime. There are some many environmental variables that symantec cant release a "best practice" because its different in every environment. Most people arent aware that if you have a scheduled continious job that when the schedule stops its still is replicating data until the outbound queue is cleared. Lots of people dont read the manual

 

[sdime808]

I've read most of the manual, is there certain section you are talking about? I have spent a lot of time reading, but I'm no VSS expert. I haven't seen anything that says you can't run a continuos job and take snapshots (if thats what you meant?). I do run a continous job but snapshots are only taken during the day, at night when the other backups are done, there are no snapshots taken. My continuos job replicates data to a different physical volume than where the snapshots are stored.

Ok and about the debate on Symantec, They built their product to be a requestor to VSS, if VVS has limitations that is going to keep if from working, they should have taken it into account. btw I'm using the assumption that this is a common problem from the info that the Tech gave me. Like when Ford sold trucks with Firestone tires that cause the trucks to crash, they did a recall. I don't want this to turn into a flame war, but I don't like how they just handed the problem off and offered almost no insight.