asked on

I/O is frozen on database []. No user action is required. However, if I/O is not resumed promptly, you could cancel the backup.

I am getting a sporadic message in my sql server logs about frozen I/O on all databases. Then about 4 seconds later, I/O resumes, and all the databases get backed up, and a dbcc checkdb is run. Why is this happening?

I have verified that our Symantec backup does not run at this time, nor does my scheduled SQL backup routine. The Symantec backup does take a snapshot prior to the scheduled backup process, but doesn't take snapshots throughout the day.

This particular server does not have a lot of users, it has an SSRS database and runs several SSIS/SSAS jobs.

Kevin Cross

My understanding is this is a function of Volume Shadow-copy Service (VSS) and tools that leverage VSS, such as the snapshot for Symantec most likely does.

EDIT: here is a write-up of how VSS backup works on SQL Server.

In addition, the following may be a useful read:
http://technet.microsoft.com/en-us/library/cc966520.aspx

jleenotlo

ASKER

I've read that too, but the I/O message is not occurring at the same time as when the shadow copy would take place, in fact, they're usually hours apart.

The Symantec procedure runs at 11pm, my SQL backup proc runs at 7:30am, and the I/O issue is sporadic. It's happened at 8am, 3pm, 11am, and all other times over the last couple days, but i've never seen the message more than twice in one day.

ASKER CERTIFIED SOLUTION

Kevin Cross

membership

This solution is only available to members.

To access this solution, you must be a member of Experts Exchange.

Start Free Trial

jleenotlo

ASKER

I've just checked and the Microsoft Software Shadow Copy Provider service is not enabled on the server in question.

Kevin Cross

Is the system virtualized?

jleenotlo

ASKER

No, it's not.

Kevin Cross

Hmm. Very strange. I am guessing you already checked SQL Agent jobs and maintenance plans to make sure their is no job getting stuck on retry during business hours, but you can double check that. I just re-read your question and see that the databases backup as well as dbcc checkdb is run. It definitely suggests this is part of a routine.

jleenotlo

ASKER

The SQL agent jobs that do maintenance (like backup) are not scheduled during business hours, and these maintenance jobs have not been failing on off hours either. I've had a few ssis jobs failing throughout the day, but they are not related to any backups.

The backup and dbcc checkdb seem to be running as a direct result of the I/O issue.

Kevin Cross

Are you seeing other messages (errors) in the log around the same time? I have seen one post suggesting issues starting after upgrade of Backup Exec, but have not seen the automatic backup/dbcc checkdb.

Kevin Cross

P.S. make sure that Windows backups are not setup as well, checking Task Scheduler also.

kevwit

What service pack to you have applied?

jleenotlo

ASKER

There are no errors immediately preceding the instances of I/O Frozen messages.

Windows backup is not even installed because we use Symantec, and the 2 of them can interfere with one another.

Windows Server 2008 R2 Standard, Service Pack 1
SQL Server 2008 R2; Build: 10.50.1777.0

kevwit

Please apply at least SP1 to that instance. You have applied CU7 to the RTM version, but SP1 fixes a lot of issues.

David Todd

Hi,

Although you're fairly certain that it isn't the backups or anything, I'd be inclined to double check that the Symantec backup using VSS isn't trying to backup the database files themselves, the mdf and ldf files..

I'd also double check the windows scheduled jobs as well.

Another thing to check is your anti-virus. Check that it is excluding the .mdf & .ldf files from its scanning and monitoring.

And lastly, do double check that the database properties, that autoclose is set to false, and autoshrink is set to false.

HTH
David

jleenotlo

ASKER

Thank you for your feedback on my question. After monitoring the systems for a about a week, I realized that there was a shadow copy starting immediately before many of the instances of I/O frozen messages. So we looked at Symantec and unchecked the options for SqlServer specific backups. This has stopped the problem for 24hrs now. So we'll continue to monitor and hopefully its all fixed.