Hey guys,
I currently have a cluster of ESXi 3.5U3 hosts I'm looking to upgrade to 4.0U1.
Reading KB1016262, it appears that my process is going to have to involve applying a patch. Since this appears to be a patch for ESX4.0, and best practise is to upgrade vCenter before upgrading ESX, the recommendation seems to be a long process:
Upgrade vCenter to 4.0
Upgrade ESXi to 4.0
Apply specific ESXi patch
Upgrade vCenter to 4.0U1
Patch ESXi
I've experienced the bug given in our test environment, and from what I understand it will not impact, for example, the host update service connecting to a server and upgrading it. Is there any reason I can't just do this:
Put hosts in maintenance mode
Upgrade vCenter to 4.0U1
Upgrade ESXi with ESXi-4.0.0-1.9.208167-upgrade-release.zip to ESXi4.0U1
It's a lot less updates and reboots, and the experience in our lab was that each upgrade/patch took quite some time, and I'd really like to minimise downtime.
PS, final step I am anticipating is using VMCLI to apply hp-esxi4.0uX-bundle-1.2.zip.
1. You don't have to put hosts to maintenance mode when upgrading vCenter.
2. You can go directly to vCenter 4u1, without 4.0.
3. Use Update Manager to upgrade ESXi's to 4u1.
4. Apply all patches that were released after 4u1.
---
MCSA, MCTS, VCP, VMware vExpert '2009
The issue with the 4.0U1 agent installation is caused by fragmented memory, so if you shutdown all vm's, reboot your hosts, then upgrade your vcenter server to 4.0U1...you should not have any issues with the vcenter agent installation. Once that is done, upgrade your hosts to 4.0U1 and you're done.
1. You don't have to put hosts to maintenance mode when upgrading vCenter.
2. You can go directly to vCenter 4u1, without 4.0.
3. Use Update Manager to upgrade ESXi's to 4u1.
4. Apply all patches that were released after 4u1.
---
MCSA, MCTS, VCP, VMware vExpert '2009
Thanks guys.
My logic here around putting the hosts in maintenance mode first was that if I was bitten by this bug and unable to connect vCenter 4U1 to my ESX hosts, there would be no issues around shutting anything down prior to upgrading the host at this point.
Anton suggests we can do one big jump, so it appears we are good to go. Thanks.