Jonathan Halliday wrote:
Brian Stansberry wrote:
> The $subject testsuite run is the second time in the last few
> testsuite runs where the process for one of the AS instances being
> tested seems to have failed. If you look at the forwarded email below,
> the testsuite fails because when it tries to stop an AS instance the
> AS isn't running.
...
> The failure occurs during the startup of a second node in the cluster.
...
> [com.arjuna.ats.arjuna.recovery.TransactionStatusManager_1] - Starting
> service com.arjuna.ats.arjuna.recovery.ActionStatusService on port 4713
>
> Jonathan, can you think of anything that goes on in there that might
> cause another AS process running on the same machine to crash?
Nope. There are various things that may stop a second instance from
coming up cleanly, but nothing I can think of that would take down an
existing one.
Thanks, good to know. Seemed a bit unlikely!
I'm mostly on vacation until later in the week, but
Michael should be able to take a look if you like. Given that the TS
version has not changed in a while though, it's either a nasty race
condition or a red herring. Is this reproducible on a hudson instance
that actually works reliably?
AFAIK, there is no such thing as one of our hudsons working reliably
lately. ;) I do know now that this portion of the the testsuite passes
on my machine. At this point we're focusing on memory issues; I'll ping
Michael if we need further help.
Thanks for responding from vacation; enjoy the rest of it.
--
Brian Stansberry
Lead, AS Clustering
JBoss, a division of Red Hat
brian.stansberry(a)redhat.com