-
Bug
-
Resolution: Unresolved
-
Medium
-
None
-
None
-
None
In this case, the controller was never downloaded, and when our infra code is
checking if it's up and running, it acknowledges that it's down and issues an
exit 1. we want that to cause the jenkins job to abort, but it's still continuing
and trying to run robot. This is confusing when looking at the jobs, but also wasting
time and $$ on resources for a job that is meaningless at that point
Here is the exact funtion that is hit in the example above. Line 1024 is the exit 1
ecelgp mentioned in IRC that we may just need to 'set -e' in the bash script.
It's probably easy to reproduce in the sandbox with any CSIT job by passing
a URL for BUNDLE_URL that is invalid. The download would fail for that case and
no controller would be installed.
I also noticed this in a different job where the failure came because a BindException
was seen in the karaf.log. That is also another reason we want to abort. just fyi
that it's not just a problem related to a single function but the entire infra scripts.