[cisco-voip] Unity 10.5.2 "promote" subscriber

[Natambu Obleton]

Thank you for the reply. I read some more on the split brain and I think I will just create a new publisher and push the db from the subscriber(utils cuc cluster renegotiate). I do have a recent backup, but I don't want the restore the weeks old voicemails to avoid more confusion that this has already started.

The reason the Publisher went down was because the VM Host was having raid array issues that was causing the storage to temporarily go away, on my other linux vm's this caused the filesystem to into read-only modes. Before we know about the array issue it looks to us like the publisher was locking up, so we left it down for a day to look into it and then it was brought up split brain, but we didn't find that out until later when users complained about MWI lights, but no voicemails. Meaning they VM huntgroup was hitting the publisher, but the subscriber was holding the voicemail and sending the MWI. So we shutdown the pub and quickly we lost 2 disks out of the raid5, but I was able to clone the VM to a new host before this happened. We fixed the VM host, but other fires burned hotter so we didn't get back to replacing the publisher until later than expected.

--

Natambu Obleton
CISSP #370414 CCIE #38491
Network Architect
FastTrack Communications, Inc.
970.828.1009

From: Ryan Huff [mailto:ryanhuff at outlook.com]
Sent: Wednesday, April 13, 2016 1:38 AM
To: Natambu Obleton
Cc: cisco-voip at puck.nether.net
Subject: Re: [cisco-voip] Unity 10.5.2 "promote" subscriber

So the quick answer is no; once an HA node, always an HA node (as far as that CUC HA pair is concerned).

Also, taking and running VM snapshots is not a supported for in UCOS (because as you mention, the longer the snapshot runs, the bigger it gets which can lead to resource issues).

At this point your best option may be to take a recent / updated DRS and restore onto a new CUC HA pair. If you don't have / can't get a recent DRS you may have to deal with COBRAS. Either way, I would make sure you have a very recent one of those, if you do not already.

What's the VM host issue that caused the primary to be shutdown in the first place? You mentioned trying to recover from a split brain; typically this issue comes about when both nodes are/were online at the same time; is this why the primary was shutdown?

Thanks,

Ryan

On Apr 12, 2016, at 11:09 PM, Natambu Obleton <nobleton at fasttrackcomm.net<mailto:nobleton at fasttrackcomm.net>> wrote:

I have run into an issue where my publisher was shutdown because of an issue with the VM Host and I have been running on the subscribe for about two weeks. I took snapshots first, then tried to do the split brain recovery, but after running the 12 hours.. it wasn't done so rolled back to my snapshots. It seems the longer I wait to recover the longer the recovery will take. Is there a way to promote the subscriber to be my new publisher and then add new subscriber for redundancy? Thanks.

​


--

Natambu Obleton
CISSP #370414 CCIE #38491
Network Architect
FastTrack Communications, Inc.
970.828.1009
_______________________________________________
cisco-voip mailing list
cisco-voip at puck.nether.net<mailto:cisco-voip at puck.nether.net>
https://puck.nether.net/mailman/listinfo/cisco-voip
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://puck.nether.net/pipermail/cisco-voip/attachments/20160413/6b7d5dd8/attachment.html>