[Buildbot-devel] buildbot master fragile with respect to outgoing smtp?

[Dan Kegel]

My buildbot master died this morning.  The log shows:

2011-08-21 09:38:34-0700 [-] Starting factory
<twisted.mail.smtp.ESMTPSenderFactory instance at 0xa10766c>
2011-08-21 09:38:34-0700 [-] releaseLocks(<BuildSlave
'example-slave-e8400', current builders: runtests>): []
2011-08-21 09:38:36-0700 [Uninitialized] SMTP Client retrying server. Retry: 5
2011-08-21 09:38:36-0700 [Uninitialized] SMTP Client retrying server. Retry: 4
2011-08-21 09:38:36-0700 [Uninitialized] SMTP Client retrying server. Retry: 3
2011-08-21 09:38:36-0700 [Uninitialized] SMTP Client retrying server. Retry: 2
2011-08-21 09:38:36-0700 [Uninitialized] SMTP Client retrying server. Retry: 1
2011-08-21 09:38:36-0700 [Uninitialized] Unhandled error in Deferred:
2011-08-21 09:38:36-0700 [Uninitialized] Unhandled Error
        Traceback (most recent call last):
        Failure: twisted.internet.error.ConnectionRefusedError:
Connection was refused by other side: 111: Connection refused.

2011-08-21 09:38:36-0700 [Uninitialized] Stopping factory
<twisted.mail.smtp.ESMTPSenderFactory instance at 0xa10766c>

It looks like my local smtp server was offline.

Is it really intended for buildbot masters to fall over dead if they
can't reach the outgoing smtp server?  I guess this is
reasonable behavior, since the smtp class that buildbot
uses probably doesn't have its own outgoing queue,
but it was still a bit of a surprise.

Thanks,
Dan