replace RAM on all cluster hosts, implications for HA ?

[xbradshr]

I have to take all HA cluster hosts down one by one and replace the RAM. They will be placed in maintenance mode individually first. What are the implications for HA. Will HA automatically elect new primary hosts, as the original primary hosts are placed in maintenance mode.

Another thing I have to watch is that my hosts are blades split across two enclosures. I need to ensure that after the upgrades, primary hosts are split across enclosures. How can I ensure this?

This is ESXi 4.1

[JohnADCO]

Even if there are a lot of VM's?  I'd be tempted to manually vmotion the VM's where I wanted them and manually vmotion them back where I wanted them after each host upgrade.

It will be interesting to read other responses on this one.

[xbradshr]

Just to clarify, I have DRS available to vmotion the VMs when placing the hosts in maintenance mode.

[JohnADCO]

I understand,  I have had HA / DRS do all sorts of unexpected things on their own is my only point when putting a running host in Maint Mode.

The VMware answer will be?   Should be fine.   Manually doing it assures you get the VM's back where you want them after / during the upgrades is all.

[a_p_]

I don't see an issue with HA while placing the hosts in maintenance mode. The second part of your question is something that actually has been modified in vSphere 5. With vSphere 4.x there's no guarantee which hosts are primaries and which are not. There is an unsupported setting that might be interesting, but it's up to you whether you set this in a production environment.

see http://www.yellow-bricks.com/2010/07/16/vsphere-4-1-ha-feature-totally-unsupported-but-too-cool/

ANdré