Comment 13 for bug 1195524

Hello!

Thank you for confirming that. I think the hostname just looked odd so just wanted to confirm as we don't have access to this environment.

If the instance is still available we really need more logs to figure out what went wrong. This will help determine if there's something wrong on the VM (disk or network issue) or a timing issue that we may be able work around.

We know the timeout set on the socket by httplib is the system default timeout for sockets, so if the following http request from ReportRoleProperties() was 'blocked' it would throw an exception on a timeout or other IO error. However, clearly the EnvMonitor was not able to complete in this case. Most likely the ifup call could not complete for some reason. If this is not a reliable operation then perhaps disabling the HostMonitor a good solution, of course the VMs other services may not be aware of a hostname change.