Hello,
I'm running a vsan 6.5 stretched cluster and I would need to put an entire site in maintenance for 1 day for a power maintenance.
Our cluster has 5 nodes on each site with a FTT=1 and a third site running the witness. I'm going to put the preferred site in maintenance, and I did not found any official doc on the best way to do it but I found a couple of post with some answers:
Doing maintenance on a Two-Node (Direct Connect) vSAN configuration - Yellow Bricks
Re: How to safely shut down one side of a 2+2 node stretched cluster
I'm planning to apply this:
- check vsan health
- change the preferred site (fault domain) to the one that will remain UP
- temporally stop DRS and manually move all vm on the remaining site/hosts (doing it manually to avoid DRS to move the vm on hosts in the same site)
- place all hosts on the site that will shutdown in maintenance mode using "ensure accessibility"
- Restart DRS
- Shutdown the hosts
Does it make sense ?
Do you know if vsan will also re-apply the FTT=1 policy and resync the vm within the same site after 1 hour ? If yes I guess it will kill the performance as we have over 60TB of used space (showing in cluster vsan capacity overview so 30TB to resync I guess ?) I currently have 108TB of free space in the whole cluster so 54TB with half nodes down, it should have enough space to re-sync locally if needed.
Thank you !
