October 7, 2003 at 1:50 pm
I am having periodic problems with my Newborn database. When running a reindexing job as part of daily maintenance, I have received the following error...
DESCRIPTION:Error: 605, Severity: 21, State: 1
Attempt to fetch logical page (1:389791) in database 'Newborn' belongs to object '0', not to object 'LAB_RESULT'.
I was able to recover with a DBCC CheckDB repair_rebuild. It happened again a week or so later, and I need to run DBCC CheckDB with_data_loss. It has continued to happen periodically. Each time the error occurs it is with a different object, most frequently an object related to the Lab_Result table. Only recently has the error been on object '0', which I understand has something to do with unallocated pages.
My initial research suggested that it was the result of a hardware error. So I replaced the RAID controller on the server. That didn't work so I moved the entire database to another server. I think that should rule out hardware.
It is still happening periodically, and I am at a loss for what to try next. Because the problem is intermittent in nature, I haven't been able to reproduce it in a testing environment, which is making it tough to diagnose.
Anyone have any suggestions?
Thanks,
Sean
October 7, 2003 at 1:59 pm
There are quite of number of KBs when searching '605' in Microsoft KB website. Have look to see whether they match with your problems.
What is SQL Server version and service pack?
October 7, 2003 at 2:07 pm
The server is Win2K sp 4. Running SQL 2000 (8.00.665)
One weird thing is that the database runs in 65 compatibility mode.
Thanks, Sean
October 7, 2003 at 2:19 pm
The database may be upgraded from SQL Server 6.5.
October 7, 2003 at 2:24 pm
Yes, that is the case, that is why it is in that mode.
All the KBs I found suggest that once the repair options are done, the problem will be gone, but with me it has recurred over and over.
October 8, 2003 at 7:51 am
It happened again last night...
DESCRIPTION:Error: 605, Severity: 21, State: 1
Attempt to fetch logical page (1:397912) in database 'Newborn' belongs to object 'SPECIMEN', not to object 'LAB_RESULT'.
Reindex job was running when the error occured. It should be noted that Lab_Result is by far the largest table in the database (10 million rows, about 2Gb in size [including indexes]).
Is up and running now, without my running anything to repair it.
October 8, 2003 at 8:28 am
Bad Hard Drives?
October 8, 2003 at 3:07 pm
I had the same problem, only in a clustered environment. I logged a call with PSS and after a long time and various tests, I received a new build (847) that provides more detailed information about system errors. You can look at the following article for more information
http://support.microsoft.com/?id=826433
Ultimately, we determined it to be driver issues and loaded the latest HP drivers, upgraded the firmware for the drives and fiber switches.
Seems to be working well now. If you haven't used the SQLIOStress utility, you may want to give it a shot. A couple notes about it though. It will affect performance on the server and you should increase the number of iterations from 5. When we ran 5 iterations, we never had a problem. But, when it hit about 15 - 18, the node would usually lock up, thus indicating a problem with the drive subsystem.
Hope this helps.
Pete
Viewing 8 posts - 1 through 7 (of 7 total)
You must be logged in to reply to this topic. Login to reply