I have a setup of 6 nodes which is running with 6.7 U1.
Today I upgraded the VCSA from 6.7 U1 to 6.7 U3.Its running fine post upgrade.I was not started ESXi host upgrade yet. https://www.virtuallysensei.com/how-to-upgrade-vsan/
After VCSA upgrade,I noticed that there was a warning on "disk rebalance"(my bad,I was not noticed this before VCSA upgrade). I triggered automatic balancing and still it is going on..I also regularly checking the status with the following command.Its going on sofar without any issues.
Actually I need to start the upgrade( with below CLI) with "Upgrade with No Downtime"
#esxcli software vib update -d /vmfs/volumes/<DATASTORE>/ESXi650-201704001.zip
One by one host upgrade-->Place the first ESXi host in maintenance mode.Select the Ensure data accessibility option.
As per our process,everyweek,I I can do 2 hosts only and it will take 3 weeks to complete the VSAN upgrade.
My doubt:-Post VSAN upgrade,Upgrade the vSAN on-disk format to version 10 is required.
When I do the VSAN on-disk format to version 10,anything I need to take care? or it will be done automatically(using Ruby vSphere Console (RVC) to continue with the disk format upgrade)? I think we can issue the command or through GUI while VMs are running.
Am I missing anything?
@ManivelR, Proactive rebalance is an intentionally slow and long-running process - this runs for 24hrs per run and isn't guaranteed to be able to rebalance all of the data in that time even.
Unless the imbalance is exceptionally severe (e.g. one host is practically empty and the others are all running at 75% or more) then I wouldn't worry too much about this and just proceed with updating the hosts.
Why are you only able to do 2 hosts a week? Unless you have to do firmware and stuff in the BIOS etc. then updating a node takes like 30 minutes and is basically just staging the patches and then a normal reboot.
Why are you using CLI and .zip as opposed to the more reliable and organised method via VUM and baselines?
Nothing needs to be manually done with regard to on-disk format upgrade - just run the pre-check, if the result is green then proceed with the upgrade - this is metadata only and should complete in a matter of minutes not hours and can be done during production hours.
Thanks for the inputs.
Why are you using CLI and .zip as opposed to the more reliable and organized method via VUM and baselines?
Ok.But we can do the VSAN upgrade either via VUM or CLI ?
Correct me please if I'm wrong.
Yes, updating vSAN nodes is basically just an ESXi update and thus VUM can be used.
VUM is also 'vSAN-aware' in that you can stage remediation for a whole cluster and it will do them one by one sequentially with Maintenance Mode 'Ensure Accessibility' (this option is used by default), or they can be done individually and just do the Maintenance Mode part yourself.