My ESXi 4.1 (Intel DH55tc motherboard, i5 and Intel add-on NIC) host keeps restarting over and over again, for seemingly no reason. It's like someone hits the "Reset" button!

The nice $upport people at VMWare told me that in order to troubleshoot that I need to get a second ESX box to send syslogs to, since every reboot wiped out the syslogs prior to that crash. (I called them because one reboot scrambled a datastore that was near and dear to my heart, and they needed 2 techs to put it back together.) On occasion I do get the PSOD, but I don't think that it's recording anywhere?

Anyway, I forgot to ask if I could use a 32-bit install of ESX 3.5 for this remote box. I don't have a spare 64-bit box for this purpose.

Can I use a 32-bit install of 3.5 for a remote syslog host?

Or, better yet, anyone have any ideas why this box started rebooting like this?

Are you ever able to get a look at the PSOD? One thing to do if you can fit it in is a memory check.

ESXi can't be used a a remote syslog receiver (I doubt ESX can do it as well) so what they may have meant is to setup a VM on a remote host to be your syslog receiver. On the VM you can setup something like Splunk. It has a free edition. You could also use vilogger from the vMA appliance, but that gets log files from a host so you might miss critical events prior to a restart.

If your datastore is usually OK, you can also set the parameter Syslog.local.datastorepath to have ESXi copy the log files to a datastore location. This duplicates /var/log/messages, but the logs will survive a reboot.