#196: Use standard logging facilities
-------------------------+--------------------------------------------------
Reporter: kparal | Owner: kparal
Type: enhancement | Status: new
Priority: minor | Milestone: 0.4.4
Component: core | Resolution:
Keywords: |
-------------------------+--------------------------------------------------
Changes (by kparal):
* priority: major => minor
* summary: Log failed job scheduling => Use standard logging facilities
* milestone: => 0.4.4
Old description:
Currently when we schedule a job and it fails, we just print it out
and
forget about it:
{{{
retval = schedule_job(control, email=email, name=testname,
required_arch=arch, dryrun=opts.dryrun,
labels=aq_labels)
if retval != 0:
print "ERROR: failed to schedule job %s" % testname
}}}
That should not be the case. We should:
* use syslog and log important stuff to /var/log/autoqa.log or similar
* send an email about failed scheduling to autoqa-results
That way we will know something is wrong and will be able to fix it.
Otherwise it just silently fails and we don't know about it.
Feel free to reassign the ticket if you want to work on it.
New description:
Currently our watchers and the main autoqa script print all debug/error
messages to standard (error) output. If it is run via cron, error output
is sent to root. But there's no log of it. We should change that. We
should use syslog and print all our messages to /var/log/autoqa.log or
similar file (we can put there a lot of debug messages which can help
resolve possible problems in the future).
So let's examine standard python logging capabilities and use them if
possible.
We don't probably need to adjust tests, because tests are executed on
clients and everything is perfectly logged and stored by autotest.
Therefore this concerns just watchers/autoqa script I think.
--
Ticket URL: <
https://fedorahosted.org/autoqa/ticket/196#comment:1>
AutoQA <
http://autoqa.fedorahosted.org>
Automated QA project