Interrupted defrag

[stonedog]

I was running a defrag on my priv.edb and it was about 60% through the 15Gb when the DOS box that the defrag was running in was shut - effectively halting the defrag.

When I restarted the defrag I started getting Jet errors. thankfully I'd taken a copy of both priv.edb and I'm now in the process of copying it back into it's orginal directory therefore overwriting the half-defragmented database.

Has anyone else ever suffered this before is their any advice you can give me.

 

[zbnet]

My advice is don't do offline defrags.

Just about the only justification for them is you just migrated a significant chunk of the users away from the server and so you want to reduce the size of the priv.edb so the backups take less time. Other than that, you don't need it.

 

[stonedog]

Thanks for the reply - managed to get everything sorted in the end. The database size was reduced from just under 16Gb to 4Gb so in this case it was worth carrying out, but I do take your point.

I just copied the original priv and pub database files back into the path (overwriting the half defragmented versions) and fired up the services.

The information store complained of a -1011 error but once I ran "isinteg -patch" to replay the log files into the database it all came back up without problems.

Ironically we're doing this housekeeping because we're moving to clustered Exchange 2003 within the next fortnight and we thought it would be a good idea to get compact and unfragmented databases

Thanks again for the reply.

 

[zbnet]

Clustered 2003? Gulp. You must be some super-duper high uptime shop, most people find the extra complexity and expense of clustering Exchange not worthwhile.

 

[stonedog]

Oh that's not the half of it! Our cluster is configured for geographic fail-over to our Business Recovery site, out mail routers are all virtual, and we haven't even implemented 2003 yet but we've started planning our 2007 migration. We literally have 24-7-365 uptime requirements to feed webmail and blackberry mobile units.

and still 70% of the mail going round our system is about nights out, football, gossip, and generally crap essentially.

ain't it always the way?

 

[fuego007]

Just so you know, running isinteg -patch does not replay the log files, it synchronizes the information store with the directory store.

It should not have been necessary to replace the databases in the event of an interrupted offline defrag, as the process creates a temporary file containing the unfragmented data. Once the defrag process is complete, the original database is deleted and the new database renamed. If the temporary database is on another drive or computer, it has to be copied to the operational server. In any case, the original database is left untouched until the defrag is complete.

Doesn't 24x7 imply 365 ?