Next ISO builds fail on Jenkins when a previous job was aborted uncleanly

We’ve seen Tails builds failing in Jenkins on isobuilder2 with huge stacktraces with the same kind of output than the one we had with #10334 (closed):

17:58:29 FATAL: hudson.remoting.RequestAbortedException: java.io.EOFException
17:58:29 hudson.remoting.RequestAbortedException: hudson.remoting.RequestAbortedException: java.io.EOFException

and same symptoms: the /tmp/tmp.xxx and /tmp/tails-build.xxx where not unmounted at some point, and then the other following builds on this isobuilder failed, because /tmp/ was full.

One thing to try if this happen again is to umount this /tmp/ directories and see if we get the same NFS filehandle error.

First build to fail seems to have been https://jenkins.tails.boum.org/view/Raw/job/build_Tails_ISO_feature-jessie/877/consoleFull (which has been deleted since feature/jessie has been merged into devel).

Then the subsequent build_Tails_ISO_devel jobs failed since
https://jenkins.tails.boum.org/view/RM/job/build_Tails_ISO_devel/1400/console

https://jenkins.tails.boum.org/computer/isobuilder2/builds can give an overview of what happened, minus the feature/jessie branch that has been removed.

Feature Branch: bugfix/10772-umount-old-builds-tmp-directories

Parent Task: #9614

Related issues

• Related to #10334 (closed)

Original created by @bertagaz on 10772 (Redmine)