Hey,
I have a SSD pool with two relative small SSDs:
Now I started to notice that one of my SSDs started to fail. So I thought, why not use this opportunity to upgrade the pool. This is how I expect it to work:
- Buy two new bigger SSDs
- Restart server in safe mode
- Remove failing SSD
- Install one of the new SSDs
- Add the new SSD to the pool
- Start the Array?
- The pool should regenerate???
- Start with Step 3 again and replace the second small SSD
- Profit ???
Is this how it works or do I really first need to use the mover and move all data back to the hard drives and replace the pool all at once?
I think this is too complicated.
Shut down all containers and VMs. Set all shares to “use cache=Yes” that otherwise are only on cache or prefer it and then invoke the mover. verify that the cache pool is empty and take the array offline or shut the server down (if you can’t hot swap) and replace the drives. start the server, assign the new drives to the cache pool, start the array and set your shares prefer and invoke the mover (this is so that shares that were set to “use cache=only” also is moved to the cache) after that is finished, set the shares to whatever it was before. Start containers and VMs. Done.
You don’t need to restart in Safe mode, you don’t need to replace each drive individually. You just do it once, move everything off, replace and move everything on again.
Personally, I have never done a replacement and upgrade of cache pool at the same time so I wouldn’t want to experiment without first having done it myself as a test before even if someone online says that it works.
But moving to the HDDs and back is pretty slow, because the cache drive contains many many small files. This is why I asked if it is possible to do it this way. Because then it would only be a ssd => ssd transfer.
You could use this opportunity to upgrade your cache pool and use ZFS in one step:
- Install two new SSDs
- configure new second cache pool with zfs
- stop all containers and VMs
- run mover
- switch all shares from old to new cache pool
- copy all data remaining data from old to new cache pool
- start containers and VMs
Thinks that’s the proper way to do it. Check the unraid forums. They have loads about this procedure.