vSAN1

hi guys, building up a POC on vSAN for the purpose of training / demo etc.

One thing I find frustrating is the apparent lack of control over it

I had an issue with one of my nodes in the cluster and while I waited for a new part I built up a lot of VM's with one of the nodes in maintenance mode. Now that my 3rd node is operational it has fulfilled the role of witness for all the disks. How can I make vSAN start moving actual data to the disk on this host? the issue I was having with the host was around storage, I want to start putting the host through its paces by being more than just a witness. Only idea I have at the moment is to put one of the other hosts into maintenance mode and chose full migration. I want more control than this though, I want to control which hosts are the component and which are the witness at a granular level - can this be done?

Stuart

There are two actions that will force a re-balance of components in a VSAN cluster. One is, as you mention, to place a host into maintenance mode (either full migration to move all components, or ensure accessibility, and VSAN will determine the components to move based on availability policy of objects). The other action that will force a re-balance is a SPBM policy change for a group of objects. But from a larger perspective, the need to granulary tune VSAN is not the intention of the platform. The concept is to set your polices and VSAN will worry about balancing the cluster in the most optimal way.  If certain capacity watermarks are hit on a per HDD level, VSAN will automatically re-balance as needed. If fine grain performance or availability control is needed, this is provided by object level SPBM policies.

ok thanks for that. Another question regarding the same thing really. I have now pulled the power on one my hosts, I am doing this as a test. I expect vSAN to start rebuilding the missing replicas on the spare host after an hour. Is there a way to see the status of a rebuild/re-mirror or whatever the correct term is?

Yes, using RVC. The RVC command vsan.resync_dashboard will show the resync status and progress of all objects. Can be used on either a per cluster or per host basis.

great thanks again, its been 1hr20mins and no re-sync has started ?

If you only have 2 active nodes in the cluster, a rebuild wont start. You need to satisfy 2n+1 = number of hosts needed to maintain fault tolerance, with n being the  FTT setting. So if you have 3 hosts and lost 1, you will still maintain availability, but a rebuild will not kick off until you are back to at least 3 hosts. If you want to have a rebuild start without having to take action to repair the host, consider (2n+1) +1.

again thanks for the reply, perhaps I should tell you what I am trying achieve and you can tell me the best way to approach:

my vsan vsan cluster started with some sas disks being marked as ssd and I now want to replace them with real SSD's. My plan was to delete a disk group at a time, allowimg vSAN to maintain data integrity by re-mirroring the data while I evict individual hosts

SSD's aren't space, they're cache. Just mark the drive bad, and replace. Unless you used them as capacity....Bearing that in mind, choose maintenance, evacuate if you can, if not stick with accessible, bring the host down, replace the drive with an SSD and bring it back online. If you only have 3 hosts, it's going to be tougher than if you had more....evacuate maintenance is the best option you have for this.