02-10-2011 12:24 AM
Hi Guys
I am busy with a new EV installation of single Mailbox Archiving server ,version 9.0 sp1
The customers wants a DR box on a VM that can be used temporarily for Shortcut Retrieval only
The Vault Store Partition is a Centera but the customer doesn't have enough storage in the DR environment to restore the Indexes in the event of USL being required
So what would the result be of the DR box taking over the Active EV server and not seeing all the Index data
I was thinking I just create the same Index Folder Structure on the DR box but not restore any Index Data and the following would happen
I know we would get errors logged but when we fail back then the Index Data would be accessible and then everything should work as before ?
Have I missed something serious that may occur if the customer doesn't restore the Index data in the case of a required Fail-Over
Solved! Go to Solution.
02-10-2011 01:28 AM
Hi Bruce
If EV sees the indexes are not there it will freak out, and try to help you by triggering rebuilds, and if 10'000 indexes are missing, it will breach the error limit and shut down EV, or if the drive it attempts to rebuild indexes runs out of space it will fail again.
So, I would set ALL indexes offline as part of the failover something like this:
USE EnterpriseVaultDirectory
UPDATE iv
SET iv.offline = 1
FROM indexvolume iv
INNERJOIN indexrootpathentry ir on ir.indexrootpathentryid = iv.indexrootpathentryid
WHEREir.indexrootpath NOT IN ('your indexes paths here')
Just remember to not allow manual archiving as well or it might cause a brain fart there too.
This way when EV goes looking for an index it will see its offline and not go all freaky on you
02-10-2011 01:28 AM
Hi Bruce
If EV sees the indexes are not there it will freak out, and try to help you by triggering rebuilds, and if 10'000 indexes are missing, it will breach the error limit and shut down EV, or if the drive it attempts to rebuild indexes runs out of space it will fail again.
So, I would set ALL indexes offline as part of the failover something like this:
USE EnterpriseVaultDirectory
UPDATE iv
SET iv.offline = 1
FROM indexvolume iv
INNERJOIN indexrootpathentry ir on ir.indexrootpathentryid = iv.indexrootpathentryid
WHEREir.indexrootpath NOT IN ('your indexes paths here')
Just remember to not allow manual archiving as well or it might cause a brain fart there too.
This way when EV goes looking for an index it will see its offline and not go all freaky on you
02-10-2011 03:47 AM
Hi Marvin
Brilliant
What about if I stopped and disabled the Index Service once Fail-Over occurs to prevent any inconsistencies ?
Or would EV already start trying to help the moment the Fail-Over occurs and USL is successful ?
02-10-2011 03:53 AM
I think EV would try to help, and if for some odd reason it starts (and the indexes are not offlined) you are boned.
02-10-2011 04:11 AM
Okay ...so then the best solution would be to take the Index offline through SQL before Fail-Over
Thx Marvin :)
02-10-2011 04:39 AM
Sorry one more idea ...you see I'm worried that the customer wont actually remember to run the SQL command in the event of fail-over
What if I don't create any Index Folders at all on the DR box ?
Then wouldn't the Index service fail when USL starts and EV would just not be able to help at all with Rebuilding Indexes ?
But the other Services would continue to run and allow users to access Shortcuts
02-10-2011 06:02 AM
I knew I had seen a post saying killing indexing was bad some where on here:
https://www-secure.symantec.com/connect/forums/disabling-indexing
The post from Wayne Humphreys is fairly definative.
Tell them as part of the USL for them (as the want no indexing in DR) is to run a SQL script, otherwise the DR is going to FUBAR