I've opened a SR on this but I wondered if anyone else is fighting this problem.
Several VI 3.5 hosts in VC 2.5 keep going into 'not responding'. Grrrr..
If I restart vmware-vpxa on the ESX host, VC looks fine again for a while, then after 5 minutes or so this all repeats again. The CPU load on the hosts that this is happening on look ok to me. If I connect VC directly to these hosts, I see nothing wrong. The VC server isn't getting clobbered either. I've had a lot of weird VC issues since upgrading to 2.5 from 2.02. I'm thinking about nuking the VC DB & Server and do a fresh build but I rather not.
I've got the log files. There's several different entries for errors. I've got a SR open with VMware now. I'll post with their findings when I get them.
We have a brand new installation of VM as well that exhibits exactly the same issues as described earlier; the hosts show (disconnected) quite frequently. After adding the IP addresses in the /etc/hosts file on the servers (including the ip for the VC server), the servers always come back within a few seconds but the time is long enough to prevent snapshots being taken and the associated VCB process to continue. Occassionally, the disconnections from VC can also result in migrations failing because the converter cannot create the virtual server on the host that has disappeared.
We are running VC 2.5.0 84767 and 3x ESX 3.5.0 826633 with the database on a virtual SQL server rather than SQL express, the three VMHosts are in a single cluster with HA, DRS and power saving enabled. I haven't opened a call with VMWare as it was suggested that the problems could be network related, but having seen the above, I think it may be worthwhile.
Andrew
I had this same problem. VMware support walked me through applying the vpx upgrade. The upgrader is located in the VC's upgrade folder we used vpx-upgrade-esx-7 linux-*
Copied it to each server
stopped the service : rpm -e mgmt-vmware
set the execution bit: chmod -x <upgrader>
installed ./<upgrader>
restarted the mgmt-vmware servers and the vmware-vpx service
All is now dandy.