Branch source builder 'Exception: out of memory'

[Paul Larson]

I was trying to run a kernel package through the system and hit this today:
21:40:46 INFO PROGRESS_TRIGGER: COMPLETED: {u'source_packages': [u'https://swift.canonistack.canonical.com/v1/AUTH_5f51954ef4b143cab042e434f7dc1524/ticket-1/linux_3.11.0-18.32~plars2ubuntu3_source.changes', u'https://swift.canonistack.canonical.com/v1/AUTH_5f51954ef4b143cab042e434f7dc1524/ticket-1/linux_3.11.0-18.32~plars2ubuntu3.tar.gz', u'https://swift.canonistack.canonical.com/v1/AUTH_5f51954ef4b143cab042e434f7dc1524/ticket-1/linux_3.11.0-18.32~plars2ubuntu3.dsc'], u'artifacts': [], u'error_message': u'Exception: out of memory', u'ppa': u'ppa:pwlars/ci-pool-001', u'exit': True, u'result': u'FAILED'}

[Andy Doan (doanac)]

Paul - is this just something where we need to specify a bigger memory constraint in juju, or do you think there's something bigger at play? I'm a little surprised - I didn't think the bsbuilder was really doing that much.

[Evan (ev)]

Do you still have this node up? Syslog and a ps listing would help get to the bottom of what spun out of control.

[Andy Doan (doanac)]

paul has a local patch that he thinks is fixing this by setting the juju constraints to 4G memory for this node

[Paul Larson (pwlars)]

No, I don't have it up anymore. I hit it when trying to run a kernel package through the system. It didn't appear to be a kernel OOM or anything like that, so it was a bit confusing. I've locally edited my juju-deployer script for bsb-worker to give it 4G, now I'm just fighting with all of our cloud options to try to find one that will let me deploy it so that I can try to go forward.