Update Manager & Loss Of VMotion Capabilities

When I discovered that there were new ESX 3.5.0 updates this morning I was happy, as I could finally try using Update Manager to patch my hosts. After I scanned my ESX 3.5.0 hosts for updates, though, my environment’s ability to VMotion stopped. Migrations would time out, and the vmkernel log showed network errors when the migration started:

Connect to <>:8000 failed: I/O error

I could ping my VMotion interfaces via vmkping, though.

The fix, for a lack of something more graceful, appears to be to destroy the virtual switch used for VMotion and recreate it. This is documented in SR#1101801771 and the tech I worked with is going to escalate it to be made into a KB article (and fixed, hopefully). If you’re having this problem try this fix, or have them look up my case notes.

It might be something with my environment, I will do more poking around to check it out. Subsequent remediations and scans seem to be working just fine now…

Update: I’ve posted a follow-up with some additional commentary, please read that along with this. Thanks!