Hola all -

We have a mix of Linux (Ubuntu & CentOS) and Windows boxes running on a Sun 4440 and a 4140; intermittently, we have issues where for an hour or more at a time, an ssh connection to one of the linux boxes (either variety) will hang for several seconds, buffering keystrokes and eventually returning. Sometimes it's a momentary pause, sometimes as long as 10+ seconds. It doesn't appear to be associated with a heavy load but it's maddeningly difficult to pin it down/test/debug appropriately. We do not ever see a similar hang with the Windows VMs through remote desktop.

We've been on XenServer since 4.5 (now 5.6) & have always seen this behavior, wondering if anyone else has seen it or has any ideas about it/how to try to debug the situation.

Many thanks for any help or ideas here.

A long shot, but we often see similar freezes when our Centos VMs power down. It causes all networking for other VMs to freeze temporarily (usually for 2-3 seconds) and then everything catches up and its fine again. Does it coincide with other VMs being shutdown?

Andy.

No, we rarely shut down our VMs; good idea though, thank you.

Yup, I see this all the time with SLES and I never could figure out why. I work around the issue by starting the following script in the background any time one of my SLES servers restarts:

#!/bin/bash
while [ 1 ] ; do ping -c 1 hostname ; sleep 2 ; done

Choose a hostname that is external to your network and is always up.

Are you running the name service cache daemon (nscd) by any chance? We've seen some occasional name resolution/network issues there, and in some cases, restarting it weekly seems to help flush out the old cache and improve response.
--Tobias

We're not running nscd here.