Otherwise the connection could have been waiting for up to 12 hours by
default. That's not fit for our purposes.
I'm trying to find out why this timed out:
http://taskotron-dev.fedoraproject.org/taskmaster/builders/x86_64/builds/6635/steps/runtask/logs/stdio
This is in Koji source code:
# set a default 12 hour connection timeout. # Some Koji operations can take a long time to return, but after 12 # hours we can assume something is seriously wrong. timeout = self.opts.setdefault('timeout', 60 * 60 * 12)
I think that a koji connection got stuck and waited for too long. I've already added some debug messages to upgradepath to help debug it next time. By configuring the timeout in koji_utils I hope we will avoid the problem completely.