I should do a blog post at some point, or you should do the MCM class 🙂

Gail - you mean you don't have all the allocation bitmap page IDs and intervals burned into your brain? Lucky you 🙂

Interesting - that Books Online page is wrong (it's based on a document I wrote while inside MS). A PFS page maps 8088 pages, not 7990. I'll get it fixed.

Skipping over all the Oracle nonsense...

There's corruption somewhere in the obfuscated-in-2005 system tables.

Can you try the following:

DBCC CHECKTABLE (65) WITH ALL_ERRORMSGS, NO_INFOMSGS

Table ID 65 is sysrowsetrefs, which is a partial...

You've got corruption in two PFS pages - this isn't repairable, even with undocumented DBCC options. You need to restore, and also figure out why this happened. Select * can...

No problem - beers are on you at PASS 🙂

That would be why the repair_rebuild failed too. You won't be able to fix this until you remove/fix the duplicate key values. Google should give you plenty of scripts to...

You should be able to manually rebuild the clustered index and it fixes it. Let me know if it doesn't and we can discuss (mostly unpalatable) options.

ok- the index really is a clustered index. Forget the MAXDOP stuff.

The errors in the word doc you attached looks like the I/O subsystem has corrupted an entire extent -...

ok - now I'm confused. Your last post says its a nonclustered primary key, which would imply the table's a heap after all, which would mean something in metadata is...

Sorry - I meant the 13 original errors. I'm concerned you might have bad memory corrupting pages before they'e written to disk.

The keys are out of order between slots 24...

It depends what the corruptions are and whether they're repairable. I think you're hitting an index rebuild bug but I need some more data from you before I can tell.

From...

Yup - there were index key order bugs fixed in SP4 (as well as tons of other stuff).

What are the keys for the table? (column name and types)

Definitely 2000. from the CHECKDB error messages.

Although this says index ID = 0, this isn't a heap - it's a clustered index. Depending on the way the indexes were built in 2000, and the way CHECKDB...