Video Screencast Help
Symantec to Separate Into Two Focused, Industry-Leading Technology Companies. Learn more.

Do i need to restart storage service regulary with EMC centera replication?

Created: 11 Feb 2013 • Updated: 01 Mar 2013 | 6 comments
This issue has been solved. See solution.

hello all,

SFW is not scanning partitions, need t0 manually restart storage service for awaiting backups to get cleared?replication seems all ok, not sure if it is EV or EMC Centera?any inputs will be valuable. Replica true, not false in dtrace.so where do I start looking at?

Enterprise vault 10.0.2

After Backup

 

Comments 6 CommentsJump to latest comment

Rob.Wilcox's picture

You should just need to put in EV in, and back-out of backup mode, and then the pending items should start to clear.

neo_hunk2kk's picture

not sure if using the EMC Centera we need to put the EV in and out of backup mode?

JesusWept3's picture

If you're usng an EMC Centera with a replica, you don't need to go in and out of backup mode at all.
When the item gets written to the Centera, it goes to the filewatch table in the Vault Store Database and then EV continually calls the Replica centera to ensure the item exists there.

Once the ClipExists() comes back True from the replica, then the item is removed from FileWatch, BackupComplete is set to 1 and any items in an "After backup" scenario are changed from pending to a shortcut (or if its a journal it deletes the item)

If your Replica Centera is down, then it will simply keep the item pending until the replica comes back up and then the items get replicated, EV Sees that the items exist and post process

If you do not have a replica centera, and you are using "After Backup", then this is a misconfiguration, because the backup will never ever happen, even if you are using NDMP to backup your centera, EV will only consider it backed up once it has been replicated, and with no replica, no backup will be considered done.

In general though, I would suggest restarting the storage every so often so that all memory and file handles are reset, as most applications have small handle or memory leaks which can mount up over time.

 

neo_hunk2kk's picture

thanks for your reply JesusWept, comnibation, After back + Replica centera

Replica centera looks to be all good, so restarting storage service is the only option i have? i was under wromg perception that dont need to restart storage service or dont need to put Ev into backup mode with EMC Centera replication. I think is somehting is right :P not sure though what it is

JesusWept3's picture

no, you don't need to restart storage or go in and out of backup mode, the centera replicates continuously and EV checks for backup continuously

If items are remaining pending then this suggests theres an issue with replication or EV checking that the clip exists.

Is Replication going through ok from Centera to Centera?
are there any nodes offline on the replica centera?
are there any DBINITS() occuring on the replica centera?

May be worth using EVCenteraVerify, put in your primary and replica centera nodes and any PEA file that you might be using and seeing what the results of that comes out to be.

But again, just to stress this, with EV Centera + Replica you do NOT need to use Backup Mode or restart storage in order for replication and backups to occur.

There is only ONE scenario where you would need to restart the storage service.
And thats if you use EMC Centera Collections...
If theres an issue writing to the primary centera or a files missing, that sort of thing, then a Centera Collector thread dies and does not get restarted.
After this happens 10 times or so, then all the collection threads will have been terminated, items will remain in pending and the Centera Collections area will grow and grow

In that situation you would restart storage and the collector threads would start up again, and the collections area would begin to shrink as the items are written to the primary centera, and as they are replicated, the items will start going from pending to full shortcuts

SOLUTION
neo_hunk2kk's picture

ClipExistsReplica (Exit) returns |True shows replica is all good.

no offline nodes

will check DBINITS(), will also use EVCenteraVerify, will verify PEA file as well.

Collections are enabled, is there any example wherein it shows collector thread dies someplace?

probably i will have to wait untill tomorow for further details from my side.

thanks for clrification that dont need to put EV in and out of backup mode or restarting storage service oftenally.