You like apples? Yeah? Well, I don’t have any. I do have a PSOD (Purple Screen of Death) for you however. While upgrading the HP BL490c blade servers to ESXi 5.1, they would spew the purple during reboots.
This is only occurring during reboots. If I cold boot the server, it completes the process and runs version 5.1 in a production environment. The behavior occurred immediately after using the HP-ESXi-5-1-0-standard-iso file in a VUM baseline. The process for upgrading changed a little during the troubleshooting, but this is the gist:
– All the HP blade serves are the same model and generation
– Upgrade the blades using the HP Service Pack iso file. I did this separately from HP’s ESXi 5.1 iso file after the initial reboot failure thinking it may have some drivers that the ESXi file didn’t. It had no impact on the PSODs. They still occurred after any reboot; a cold boot resulted in ESXi loading and running.
– Patch ESXi 5.0 using VUM prior to running the 5.1 upgrade. This also made no difference in the behavior.
I contacted VMware support on this problem. It took a couple days, then verified that 5.1. Update 1 fixes the problem. That didn’t help me much as HP hasn’t released U1 for their ESXi installation. VMware and HP are currently talking to get an update to the 5.1 installation iso.
If any HP Blade admins are running into this, a fresh installation of 5.1 works fine. I figured I’d try it, since nothing else worked. I used the same HP iso file that failed (sort of) during the upgrade. So, it must be something in the upgrade from HP and how it interacts with vSphere. The other hint that it may be the upgrade file is the “PF Exception 14” in the PSOD, which indicates a hardware problem.
If you do a fresh install in a fairly large environment, make sure you have an up to date host profile to apply to the “new” host, so you don’t have to recreate network connections, storage adapters, etc.
I plan on giving VMware and HP a couple days to figure this out before I start doing fresh installs on the hosts.