We recently had an outage on our DC – one disk out of a 4 disk raid array failed and it ran in degraded mode for about 3 weeks before we were able to replace the disk. By the last day before it got replaced, exchange was running so poorly that for some users it would fall over as soon as they opened their mailbox (I assumed they had corrupt mailboxes). We replaced the disk and the rebuild was a success, but when the exchange DB was still corrupt, and got restored from backups a week only – blowing away a small subset of emails.
All that was completed by another staffmember, but here is where I step in – about a month later and some staff are discovering that an email that was lost is super important. When they restored from the backup they put a copy of the exchange database they restored to in a new iSCSI drive along with the logs from between the backup being taken and the server being restored. From what I have been told, this should be enough to get us a working view of the mailstore on that day so that I can get the 5 or 10 really important messages out of there.
Problem is I know nothing about exchange. Google suggested http://technet.microsoft.com/en-us/library/bb124040(v=exchg.65).aspx and similar pages but these seem to be to do with repairing the running database, which is not what I want – right now email is working, I don't want to change that.
We use SBS 2008 and Exchange 2007 – any suggestions on where to start would be greatly appreciated.
Best Answer
At this point in time, if the emails are "super important" then I'd strongly suggest getting someone who does know exchange fully involved. I don't want to sound disrespectful, but this is not the best point to start learning from.
If you can't do that (though I really do think you should try - this situation sounds like it would benefit from an on-site look-over from an experienced exchange consultant), I'd suggest using out of band recovery tools - something that allows you to mount and inspect the information store outside of exchange. I use ontrack powercontrols when I help people to do this DR route, but there are other options out there.
How did you/whoever establish that exchange had a corrupted datastore anyway? That shouldn't happen because of one disk in a RAID array dropping out - either there was no fault (or at least no corrupt database issue) with the Exchange server or you still have an issue with this system that you need to discover and nail down before it goes wrong again.