Hi all
I am looking to migrate from a single physical EV journaling server (EV 10.04) to a distributed virtual 4 server model, with 2 servers running joural tasks and storage services, and two servers in an index server group. I'm familiar enough with the reason for/against index groups, but we do use CA/DA and the single server is maxed out - if all is ok it copes (just) but any issues with corrupt volumes, SQL or other problems, it can take days to catch up, hence the distributed model plan.
My question here is the strategy for moving to this model, for the indexes. It would appear this is the correct process:
- install new servers with indexing services only (and task controller for index admin task)
- create new locations on each server
- create new index server group with the two new servers
- assign the journal vault store(s) to the new index group
- close the old index volume locations on the old server
At this point, as far as I can tell, the new servers will take over indexing for any new items or index rebuilds, and the old server will continue to handle searches on the old indexes. However I still need to move the index locations to the new servers so I can decommission the old server (I'm ignoring storage and tasks, the process for them is more straightforward).
This howto article http://www.symantec.com/docs/HOWTO82325 tells me that I cannot move from an ungrouped server to a grouped server, so I guess I will need to add the old server to the index server group, follow the howto article, and then remove the server from the group. However I have read in several places that once indexing has begun you cannot remove servers from a group. Maybe this does not apply to the old server in my situation, since the locations it manages are closed, and it will not have had any items indexed by it while in the group.
So can anyone please advise if they have worked through this process, and will I be able to successfully remove the old server, without losing the several hundred million indexed items it managed? Or is there a different / better way to manage this? I am attempting to work through this in the lab but I am very nervous going into production :).
Any advice, comments much appreciated.
thanks, Andrew