I'm using ASE 12.5 with EBF 11442 patch. I keep encountering 605 errors. There was no hardware issues involved. The tables involved are using unique cluster index. I run dbcc checktable and tablealloc and found no problems. I am only able to resolve this by creating new copies of the tables. Anyone have encountered this?
Originally posted by engkai
I run dbcc checktable and tablealloc and found no problems. I am only able to resolve this by creating new copies of the tables. Anyone have encountered this?
605 is a hard fault. None of the dbcc's will fix this.
By regularly creating new tables, you are only delaying the occurance of 605 until you hit it again. checkdb should locate 605.
If you have located the table that is reporting 605 and isolated that hosting device. Then you could carry out a salvage operation by moving all the objects on that device to a new device and completing quartine that device in the database by changing the segmap to 0. This only if there aren't any clean backups available, otherwise going backup is the best option and apply the tran logs since then to bring the database to current.
I have a 605 Error one year ago. This was an hardware error. I had a message like this:
00:00000:00012:2003/03/25 11:33:45.01 server End diagnostics for read failure:
00:00000:00012:2003/03/25 11:33:45.01 server Run the DBCC command 'dbcc tablealloc' for both the objects 99, 3
00:00000:00012:2003/03/25 11:33:45.01 server Error: 605, Severity: 21, State: 1
00:00000:00012:2003/03/25 11:33:45.01 server An attempt was made to fetch logical page '298449' in database 'sales_db' from cache 'default data cache'. Page belongs to object with id '3', not to object ''.
This kind of error are very difficulty to resolve. I can fix the error touching one OAM page that was wrong, but this is out of any rule and recommendation. The problem was an OAM Page (allocation page) that change his obid = 99 for 3, that mean that just one bit change the "pageno" into the page header, and the OAM pages start to be a page of the syscolumns. I was lucky because the allocation page was clean.
Before do that I study very deep the pages that had the error, and the tables.
You need to find the page with the error and find inside how many error there are. The hardware error could change everything. You will need use "dbcc allocdump" and "dbcc page".
"The recommendation in this case is rebuild all the tables (BCP) in a new database."
If you have a problem with an OAM pages, then the database will not grow any more :-(. If you have the problem just with table pages then droping the tables may be fix the error.