I had a lot of trouble reproducing this, but as it seems to mysteriously pass when run 2 times in a short lapse, I traced it to potential race conditions in tempfiles.
Unlike xenial, both precise and trusty jenkins nodes have a large amount of breadcrumbs in /tmp (200,000+).
I had a lot of trouble reproducing this, but as it seems to mysteriously pass when run 2 times in a short lapse, I traced it to potential race conditions in tempfiles.
Unlike xenial, both precise and trusty jenkins nodes have a large amount of breadcrumbs in /tmp (200,000+).
ubuntu@ juju-lcy02- machine- 1:~$ find /tmp -name 'tmp??????' -type f -user ubuntu |wc -l
213805
I'm assuming this causes just enough slowdown to randomly fail, so I went ahead and did a cleanup of those old temp files.