1615312 - generic-worker: simple engine does not always clean up task directories

Status: RESOLVED FIXED

(Taskcluster :: Workers, defect)

Description

[Pete Moore [:pmoore][:pete]]

Currently the rasbperry pi CI worker runs out of disk space from time-to-time since old task directories are not always getting cleaned up.

On closer inspection, rm -rf <task dir> fails sometimes since some of the files have had their user-writable bit unset.

This can be fixed by running chmod -R u+w <dir> before running rm -rf <task dir>.

At the same time as implementing the change, it would be worthwhile to use the native chmod and rm commands on the host, since these produce useful output in the case of failure.

Comment 1

[Dustin J. Mitchell [:dustin] (he/him)]

https://github.com/taskcluster/taskcluster/pull/2347

Comment 2

[Pete Moore [:pmoore][:pete]]

Released in taskcluster 25.3.0.