October 16, 2009 at 7:23 am
Let me give you a run-down of the system so I can try to pinpoint the problem.
I have a Dell Blade Server connected to a Dell SAN. The connection uses Microsoft iSCSI Initiator with Dell drivers.
I have ran diagnostics on the Disk and Memory and both checkout fine.
The HDD space allocated is 1.90T. Of that 1.90T there is 1.2T of available space. And the machine has 16GB of RAM. I have set SQL Server to enable large page extensions since this is on a 64-bit machine.
I have 3 databases on this drive.
1 - 223GB
2 - 173GB
3 - 204GB
Plus the tempdb is located on this drive.
Totaling 600GB. These are transaction type databases in the fact that we have users connecting remotely through a program and updating these databases almost non-stop.
We output data from these databases quite often so we create a snapshot. The only database that we take a snapshot of is number 2 but we have to take 2 snapshots, one for each data output we need to create. These snapshots sit on the HDD until the next time we need to take one which is often for 1 but could be once a week or so for the other.
Every once in a while, more often now then before, we get the error insuffiencent resources and the database pretty much freezes and we can't do any transactions. Even a reboot doesn't help, the machine comes back up and the database is in recovery mode and the second the recovery is complete, the i/o error occurs again.
To get it back working, we have to drop the snapshots (which freezes) and reboot the system. The database goes into recovery mode and when finished, everything works ok.
Any ideas? I have attached the SQL boot log, the one that gets the errors.
October 16, 2009 at 8:17 am
You could check http://support.microsoft.com/default.aspx?scid=kb;en-us;304101, but i'm not convinced this will help with your issue.
Something is being maxed out. You could try running some performance monitoring for CPU, disk, memory etc. PAL (www.codeplex.com/PAL) can be really helpfull here.
My guess would be that Average Disk Queue Length will be high (I think 2+ per disk is considered high).
The solution 'may' be to setup smaller LUNs (viewed as physical drives in Windows) on your SAN and separate Data files from T Log files and Snapshots. Putting Tempdb on another one may be helpfull too.
We have one LUN fro data files, one for TLogs and another for Snapshots. It seems to work quite well.
Viewing 2 posts - 1 through 1 (of 1 total)
You must be logged in to reply to this topic. Login to reply