November 6, 2005 at 10:46 am
I am running win2003 EE os with sql2k sp4 on a Dell 2680 with 12 gb ram and 3.2 tb of raid 5 build space.
I had just finished building a 1.2 tb delivery db when I started to shrink it to fit on our delivery server (220gb of free space needed to be removed.). I let the shrink run for 14 hours, but it was hung. SP_who2 showed the spid was blocked by itself. I dont know how that can happen without triggering a deadlock scenario, but I killed the spid. When I checked back with sp_who2, I saw the spid was gone, no rollback, so I rebooted the server to be safe.
After the reboot, the db was in recovery mode. It took 4 hours to get past the analysis mode, but when it got to 99%, I received a bunch of 17883 errors in the app log and sql errorlog complaining about "Process 10:0 (310) UMS Context 0x002a86c0 appears to be un-yeilding on scheduler 1". This went on for about an hour when the problem seemed to clear itself up somehow and now the db is in a reeeaaal slow recovery mode.
All of the threads I could find spoke about 17883 errors coming from queries, but nothing about recovering databases. Does anyone know how a spid can block itself? I just applied sp4, but I am about to roll it back and apply sp3 instead. I havent had issues with this server until sp4 came along.
has anyone else had a simular experience?
Thanks!!
November 7, 2005 at 7:00 am
hi.
sorry to hear about your troubles but you shuldn't have killed the process. In sp4 it's OK to see the processes locking themselves
look at this KB
http://support.microsoft.com/default.aspx/kb/906344
Yoav.
Viewing 2 posts - 1 through 1 (of 1 total)
You must be logged in to reply to this topic. Login to reply