You upgraded the hosts before vCenter itself? You should have done it the other way around, the supported way..
Never run newer hosts with older vCenter, you ask for trouble then..
So, the procedure to upgrade from vSphere 4.o to vSphere 4.0u1 is:
upgrade host with VUM
upgrade VmWare tools on the virtual machine
Is this ok?
Maybe I'm forgetting something?
There are some operation for VCB, Converter or any other plugin?
Sorry for the question, but I'm more frightened by this operation rather than the migration from 3.5 to 4.0 (in that case i found a lot of docs that helps a lot).
Essentially that is correct.
1. Upgrade Vcenter Server(s) first (you should do any that are linked mode together)
2. Upgrade all hosts managed by that vcenter server
3. Upgrade VMware Tools
4. Upgrade VM Hardware after VMWare tools.
You should read through the release notes here:
You should read the Upgrade Checklist here:
You should read the ugprade guide here:
When you read the upgrade guide, read the sections that apply to you, as it includes most upgrade scenerios.
after upgrading my LAB with 4 * DELL 6850, 3 of them didn't come back after the reboot.
After rebooting all servers again, one server successfully passes the store-logs sequence, two servers still didn't pass the store-logs sequence even after several reboots.
Would appreciate any help or hint.
Greetings from germany.
To clarify the last doubt: I'm going to upgrade from vSphere 4.0, so i think that the virtual hardware upgrade shouldn't be necessary (I upgrade the hardware from version 4 to version 7 during the upgrade from esx 3.5 to vsphere 4.0).
Is this right?
I ran the pre-upgrade tool and i noticed a strange behaviour: the tools reports ok for all my 4 hosts, but in the host selection windows it reports a wrong software version for my hosts:
I'm sure that all my host are running the same version (4.0.0 build 175625), as displayed in the host summary tab in VIC.
The tools recognize 2 hosts as esx 3.5.0 build 163429, one 4.0.0 build 175625 and the last one 4.0.0 build 164009....
I really don't understand these results, but, as I already said, the final report tells me that there should be no problem in upgrading all of my hosts...
What should I do?
I continue the vcenter upgrade process or i stop the operation?
I updated all three hosts, then updated vcenter.
Then updated vcenter's vmware tools. Everything reports build 208167. Also the false positives about VM's running too much memory or CPU have gone away finally (nehalem cpu bug).
The only issue is one of my vm hosts lost redundancy to the SAN. The second nic which is a bnx2 (Broadcom NetXtreme II BCM5709) will not connect at gigabit to the switch. It will only negotiate 100mb, and VM's are god awful slow when it's set to that. If I force it to 1000mb full duplex, that nic just drops off. So we have 3 IDENTICAL servers (Dell R710's), and only this one is having the issue. Next I'm going to check cabling (it's a prefab cable, not in house crimped). The Cisco 3750g switch it connects to shows frame errors on that port.
Strange it didn't happen until the upgrade. Related, or coincidence? hmm...
The only thing I noticed is that when upgrading each host, I guess it takes so long for them to come back up that it times out and you have to manually have to tell vcenter to reconnect to that host. When it does this it says "Wake on Lan is not supported by this NIC". However that's false. I can still shutdown, startup and standby vm hosts.
In response to my post... it was a weird coincidence. Wiggling on the cable at the network jack on the server end restored 1000mbps full duplex link and the connection to the SAN is now redundant.
That'd be a bad cable ... but check just the same that you're plugged into the same ports on all the servers.
We just got some servers with 4 NICs on the motherboard, and found that the first two are high-end intel NICs with TOE, and the second two NICs are cheaper, without TOE...
Sorry, I don't want to annoy anyone,but.... no one else encountered the behaviour that i explained in my precedent post?
Can I proceed with the upgrade process even if the pre-upgrade check utility shows me different software version for my esx hosts?
just to confirm that, The A specified parameter was not correc" error: spec.memAllocation.overheadLimit problem is still exists in vSphere 4-U1.
I've just upgraded my hosts to vSphere U1, after i found everything working fine, i have shutdown the vCenter VM to upgrade the VM-Tools, and accessed to the host where the vCenter was running. when i ugraded the VMTools, and PowerOn the vCenter VM, i see the error still appearing in the host.
A specified parameter was not correc" error: spec.memAllocation.overheadLimit
Other than that, it seems everything fine on two hosts. Will wait and see what other issues might come before i upgrade my other hosts on the Dev & Test environment.
Hussain Al Sayed
If you find this information useful, please award points for "correct" or "helpful".
ESX 4.0 Update 1 has been temporarily removed for download. Looks like there is a serious issue with HP management agents when updating hosts -->
I have done it a few times now at different sites. No problem.
1) vCenter to 4.0 - About 5 minutes
2) Each host (ESXi 3.5 -> 4.0) about 25 mins / host
3) New tools (1 minute and a reboot)
4) New virtual hardware (Shut down VM - New vHardware - Startup VM again - Finds new hardware - Reboot)
Repeat step 3 and 4 per VM when OK to reboot that VM - Or automate and do them all at once if you can.
1) Is still true for 4U1. Did it today.
2) I haven't tried yet. Client was on fat ESX 3.5 and wanted to go to ESXi 4U1 so I did a clean install on all hosts. But I would guess the update is more or less the same time. I wonder why it takes it almost half an hour (ESXi4U0 anyway)? Ah well...
Whats that New V Hardware step your indicating? That only applies when upgrading from ESX 3.5 to 4.0 U1 right? I was already on 4.0 build 17xxx. I don't have to do that right? I don't even have a "New Hardware" button. Just an add or remove button. All my VM's are VMware version 7.