Is it possible to take one target on one HA node offline?

[jhamm@logos-data.com]

I have an HA cluster with 2 nodes and about 10 targets. Each target is a RAID 1 disk array. I have a predictive failure on one of my hard drives on one of the nodes. Is it possible to take this one target on this one HA node offline that has the bad drive, while leaving the target available to clients via the other HA node? I would like to take this one target offline while I replace the drive and rebuild the RAID 1 array, and then sync it and bring it back online once the rebuild has completed.

Make sense? Possible?

Thanks,
Jeff

[Max (staff)]

That's really easy with HA - right click on the HA device that will remain online, select replication manager.

RepMan01.jpg (22.36 KiB) Viewed 6343 times

Remove the replica

repman02.jpg (42.49 KiB) Viewed 6344 times

Now go to the node you'll be turning off and delete the HA target you just excluded in the replication manager. Stop the StarWind service. Now you can do all the necessary actions with the disks.

repman03.jpg (29.21 KiB) Viewed 6343 times

When the server is back online - do the necessary initialize-format-assign letter staff. Now check if StarWind service has already started. Go back to the replication manager and press the "add replica" button.

repman04.jpg (26.38 KiB) Viewed 6344 times

Point the wizard to the partner server

repman05.jpg (36.45 KiB) Viewed 6343 times

Choose the image file(if left) or enable the "create new" checkbox

RepMan06.jpg (44.47 KiB) Viewed 6345 times

Select the sync and heartbeat channels.

RepMan07.jpg (35.3 KiB) Viewed 6346 times

Select the previously defined cache size for the HA device. Do not increase or shrink the cache!
Complete the wizard steps and wait for the HA device to finish the synchronization. Now you can reconnect the target back to the client.

[jeddyatcc]

I'm just curious though, if it is a RAID 1, no downtime is required at all for that node while it is rebuilding. Modern RAID controllers will rebuild a RAID 1 very quickly. If you have to take the whole server down(no hot swap), then I would just shut it down replace the drive, rebuild the array inside the RAID card console and reboot. Once the machine is up StarWind will resync and enable the target.

[jhamm@logos-data.com]

jeddyatcc - It is a a hot swap RAID 1 set, and so it can rebuild online. But my thinking was that during the rebuild the performance would be lower, so it might be better to just take that one target offline for performance reasons during the rebuild, and direct all clients to the other HA node for that target until the rebuild process completes

Max - great instructions, and you've maybe given me an idea for another thread I had opened about "Change cache memory size on HA target "

Can I use these same instructions (except the part about taking the StarWind service offline) to increase the cache size on my HA targets with zero downtime?

Thanks,
Jeff

[jeddyatcc]

jahamm

Performance would be slower during the rebuild, but not much. Also, depending on how large the drive is and what type, the rebuild can take quite some time (typically 80MB/s maybe a little higher so 3TB drive = 11 hours). Letting the RAID controller rebuild from the console with nothing else to do would probably be the best idea. BTW RAID 1 with anything larger than 1TB is asking for trouble as you can have rebuild problems. Your other option would be to completely destroy the RAID and recreate it with the new drive, as you will have to perform a full Sync either way.

I replied on the other thread as well, as yes I think that is possible.

[Max (staff)]

Jeff,
Although this option is not greyed out it's not actually working for both nodes. So I won't recommend changing cache size this way.