Please start any new threads on our new
site at https://forums.sqlteam.com. We've got lots of great SQL Server
experts to answer whatever question you can come up with.
| Author |
Topic |
|
rkirk
Starting Member
10 Posts |
 Posted - 2005-08-10 : 15:12:24
|
| We are seeing random 605 Severity 21 Errors. We applied SP4 over the weekend, but sure enough Monday night we received another one.We repaired the DB and don't have a copy of it when it was correct, so I can't send my DBCC results. I do have this from the Errorlog though. It occurred during our database load process.2005-08-08 19:39:25.95 spid77 Getpage: bstat=0x9, sstat=0, cache2005-08-08 19:39:25.95 spid77 pageno is/should be: objid is/should be:2005-08-08 19:39:25.95 spid77 (1:1787)/(1:1787) 610101214/12661035512005-08-08 19:39:25.95 spid77 ... IAM indicates that page is allocated to this object2005-08-08 19:39:25.97 spid77 Getpage: bstat=0x9, sstat=0, cache2005-08-08 19:39:25.97 spid77 pageno is/should be: objid is/should be:2005-08-08 19:39:25.97 spid77 (1:1787)/(1:1787) 610101214/12661035512005-08-08 19:39:25.97 spid77 ... IAM indicates that page is allocated to this object2005-08-08 19:39:25.97 spid77 Getpage: bstat=0x9, sstat=0, cache2005-08-08 19:39:25.97 spid77 pageno is/should be: objid is/should be:2005-08-08 19:39:25.97 spid77 (1:1787)/(1:1787) 610101214/12661035512005-08-08 19:39:25.97 spid77 ... IAM indicates that page is allocated to this object2005-08-08 19:39:25.97 spid77 Error: 605, Severity: 21, State: 12005-08-08 19:39:25.97 spid77 Attempt to fetch logical page (1:1787) in database 'db1' belongs to object 'IEFPROD_FACA_GEO_LOC', not to object 'FacilityInterest'.. |
|
|
paulrandal
Yak with Vast SQL Skills
899 Posts |
Posted - 2005-08-10 : 15:25:51
|
| Yep - its corruption, but we're stuck without the CHECKDB results.Paul RandalDev Lead, Microsoft SQL Server Storage Engine(Legalese: This posting is provided "AS IS" with no warranties, and confers no rights.) |
 |
|
|
rkirk
Starting Member
10 Posts |
Posted - 2005-08-10 : 17:08:42
|
| I restored a backup that was taken while the database was in it's corrupt state, but the restored version comes up clean.In the future what is the best way to save the database in it's corrupt state so that I can repair it and make things available and also have information for debugging later?I'll at least save all of my dbcc outputs next time for posting.Thanks! |
|
|
|
paulrandal
Yak with Vast SQL Skills
899 Posts |
Posted - 2005-08-10 : 18:27:44
|
quote:
Originally posted by rkirk
I restored a backup that was taken while the database was in it's corrupt state, but the restored version comes up clean.In the future what is the best way to save the database in it's corrupt state so that I can repair it and make things available and also have information for debugging later?I'll at least save all of my dbcc outputs next time for posting.Thanks!
Taking a backup is the only thing I can think of - ah - what may work even better to preserve corruption is to shutdown SQL Server, take physical copies of the db files and then restart. This has the obvious downside of having to cycle the server though.What you may have seen is that the on-disk database was perfectly fine, but a page in the buffer pool was corrupt and CHECKDB didn't flush out enough of the buffer pool to cause the page to be read in from disk. I've seen this happen several times while investigating corruption that's turned out to be caused by bad memory.ThanksPaul RandalDev Lead, Microsoft SQL Server Storage Engine(Legalese: This posting is provided "AS IS" with no warranties, and confers no rights.) |
|
|
|
antstanley
Starting Member
1 Post |
Posted - 2005-09-08 : 09:56:31
|
| Hi there,Got the same problem on 2 servers here, one of which I manage. Did a restore from a backup repaired it. It comes out. But within a few days it comes back. And I have to do the whole process again. Backup, Restore, Repair. Luckily this database is not a production database, but we need this resolved. The 2 servers giving the same problems are both running Windows 2003 and SQL 2000, with different service packs on each. They are both HP machines with the same spec and sit in the same physical rack. We do have other databases running Win2003 and SQL2000 but on IBM machines no problem.Whats up |
|
|
|
ryanston
Microsoft SQL Server Product Team
89 Posts |
Posted - 2005-09-08 : 13:34:40
|
quote:
Originally posted by antstanley
Hi there,Got the same problem on 2 servers here, one of which I manage. Did a restore from a backup repaired it. It comes out. But within a few days it comes back. And I have to do the whole process again. Backup, Restore, Repair. Luckily this database is not a production database, but we need this resolved. The 2 servers giving the same problems are both running Windows 2003 and SQL 2000, with different service packs on each. They are both HP machines with the same spec and sit in the same physical rack. We do have other databases running Win2003 and SQL2000 but on IBM machines no problem.Whats up
Can you please start a new thread in this forum, and the output of DBCC CHECKDB (<your_database>) with no_infomsgs, all_errormsgs?Thanks,----------------------Ryan StonecipherDeveloper, Microsoft SQL Server Storage Engine, DBCC(Legalese: This posting is provided "AS IS" with no warranties, and confers no rights.) |
|
|
|
Westley
Posting Yak Master
229 Posts |
Posted - 2005-09-12 : 02:09:48
|
| Guys, I've got the same issue here, just wondering how often does it happening to u guys? We contracted MS and needs to run SQLIOStress test on the box, still waiting from the business to give us the downtime since its a production box :(If you guys got other ideas which helps, pls let me know. |
|
|
|
ryanston
Microsoft SQL Server Product Team
89 Posts |
Posted - 2005-09-12 : 12:04:54
|
| These errors are typically caused by some sort of database corruption, either in-memory on on-disk. Essentially, the server is requesting a page from the buffer pool, and the page number/object ID on the page differs from the one that was requested. When these values are garbage (i.e., do not reflect any existing object in the database), then you can almost always chalk it up to either bad hardware, or an in-memory scribbler. If they're legitimate object IDs, then something else weird is going on (which could still involve hardware), and you'll need the help of PSS to diagnose the problem. (Typically, they'll analyse the transaction log, look at Profiler traces, run SQLIOStress, etc.)If you want, I can take a quick look at your CHECKDB output. Please start a new thread with the results of DBCC CHECKDB ('your database') WITH NO_INFOMSGS, ALL_ERRORMSGS. Thanks,----------------------Ryan StonecipherDeveloper, Microsoft SQL Server Storage Engine, DBCC(Legalese: This posting is provided "AS IS" with no warranties, and confers no rights.) |
|
|
|
|
|
|
|
|
605 Severity 21 Errors