November 22, 2004 at 4:20 pm
Last week I started receiving the following error every 15 minutes, it came from different users across the network
“The Kerberos subsystem encountered a PAC verification failure…”
Did a bit of research, it looked like some changes the network admin had been making to our firewalls was causing it. I was going to wait until they could be rebooted to see if the error went away. At this point it was just an annoyance as the user with the error could connect 2 seconds later.
Within 2 hours I had more errors than I knew what to do with. DCom errors, the MSDTC had crashed, nothing remote would work expect EM. I couldn’t even bring up the start menu on the box itself. Everything gave an insufficient resource error. Fine, Shut down SQL server from another sql box and was finally able to gain a bit of control over the box and shut it down safely.
It came back up fine and I have not encountered the errors again, however my backups are taking an hour longer now. Every step (full backup, reindex, tlog backup, robocopy to backup server) have all increased. The databases have not grown, there are no hardware errors reported. I can’t find any rhyme or reason for this change. I rebooted it again thinking that maybe something was not quite cleared up but it’s the same.
Server specs: Server 2003, SQL 2000
If anyone has ideas as to what else I could look at it’s greatly appreciated. I need to find a way to get these back down
Thanks
K
November 22, 2004 at 5:21 pm
Are you performing the backup locally and then using robocopy to transfer it accross the network. If so, has the backup slowed or just the transfer. If the copy, can you try writting to another network drive and note the performance? Have the network guys turned on any hardware or network level compression?
November 22, 2004 at 5:28 pm
It backups locally then Robocopies to a network drive. The frustrating part is that the backup step slowed down 17 minutes and the robocopy step slowed down 16 minutes. SO both the local stuff and the network stuff is slower.
It almost seems like a disk problem because both reads and writes are slower but when I check them everything looks fine.
I double checked and there is no compression on the network drive and certainly not on the SQL server.
November 23, 2004 at 3:04 pm
Definitely sounds like a hardware issue associated with the drive you are reading/writting from. Is it possible to try writting to a different local drive on a completely different drive controller? Perhaps simply install a new one even if it is only for testing purposes. It's tough to tell if the problem is with the actual disk or the controller (raid?) but it sounds like one of the other is the root cause of the problem.
November 30, 2004 at 3:26 pm
Well it happened again. Did some more digging into the specific services that were failing and I'm sure it'a a low memory condition causing such havoc.
I've confirmed with Imceda that there is memory leak in LiteSpeed version 3.2.1.16 but that it is fixed in version 3.2.1.23. Let's hope so.
I've started the upgrade process, time will tell I guess.
Thanks for your suggestions
Kelsey
Viewing 5 posts - 1 through 4 (of 4 total)
You must be logged in to reply to this topic. Login to reply