Hi there,
it seems that failed jobs will never be executed again.
My extension creates jobs that talk with an external server via soap. This connection might fail (e.g. server maintenance) which then means that my job fails and needs to be reexecuted later. On my wiki I am running $wgJobRunRate = 0, my extension job is in $wgJobTypesExcludedFromDefaultQueue. To run my jobs I have a cronjob with "php runJobs.php --maxjobs 10 --type CreateTicket". My job has allowRetries() implemented with "return true".
From my observations the "job_token" and "job_token_timestamp" fields in the job table are used as a LOCK so that no job gets executed twice. The only problem is that these fields are never emptied, if the job fails (return false), so that the job scheduler can reexecute them.
Is this intended behaviour or is this a bug? Does anyone have an idea how to solve this issue?
Thank you. Regards Henning Vreyborg
Hello,
I want to have my job reexecuted after it fails. Is my solution of adding a clone to the job queue viable? For me it feels a bit dirty.
class CreateTicketJob extends Job { [...] public function run( ) { [...]
if (/* job failed due to not connecting to external resource */) { $job = new CreateTicketJob( $this->getTitle(), $this->params );
return JobQueueGroup::singleton()->push( $job ); } return true; } [...] }
Any help is appreciated.
Sincerely Henning
Am 02.07.2015 um 14:51 schrieb Henning Vreyborg:
Hi there,
it seems that failed jobs will never be executed again.
My extension creates jobs that talk with an external server via soap. This connection might fail (e.g. server maintenance) which then means that my job fails and needs to be reexecuted later. On my wiki I am running $wgJobRunRate = 0, my extension job is in $wgJobTypesExcludedFromDefaultQueue. To run my jobs I have a cronjob with "php runJobs.php --maxjobs 10 --type CreateTicket". My job has allowRetries() implemented with "return true".
From my observations the "job_token" and "job_token_timestamp" fields in the job table are used as a LOCK so that no job gets executed twice. The only problem is that these fields are never emptied, if the job fails (return false), so that the job scheduler can reexecute them.
Is this intended behaviour or is this a bug? Does anyone have an idea how to solve this issue?
Thank you. Regards Henning Vreyborg _______________________________________________ Wikitech-l mailing list Wikitech-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikitech-l
On Tue, Jul 7, 2015 at 10:28 AM, Henning Vreyborg < marshen@mail.uni-paderborn.de> wrote:
I want to have my job reexecuted after it fails. Is my solution of adding a clone to the job queue viable? For me it feels a bit dirty.
class CreateTicketJob extends Job { [...] public function run( ) { [...]
if (/* job failed due to not connecting to externalresource */) { $job = new CreateTicketJob( $this->getTitle(), $this->params );
return JobQueueGroup::singleton()->push( $job ); } return true; } [...]}
That's how GWToolset https://www.mediawiki.org/wiki/Extension:GWToolset does it (for a different purpose). It is indeed dirty, but it works. It makes the job hard to kill though, so you might want to build in some safety mechanism to avoid involuntary DOS attacks.
Have you tried setting something like:
$wgJobTypeConf['default']['claimTTL'] = 3600;
Jobs are not normally retried by default, only archived and deleted...maybe that default should change.
-- View this message in context: http://wikimedia.7.x6.nabble.com/Failed-jobs-will-never-run-again-tp5049689p... Sent from the Wikipedia Developers mailing list archive at Nabble.com.
Hi,
yes this works but only together with 'maxTries' which is by default 3. But even if I set this to something very big it still might happen that a job gets deleted. In my case I can't risk that any job gets deleted. So I'll stick with cloning.
Regards Henning
Am 08.07.2015 um 11:50 schrieb Aaron Schulz:
Have you tried setting something like:
$wgJobTypeConf['default']['claimTTL'] = 3600;
Jobs are not normally retried by default, only archived and deleted...maybe that default should change.
-- View this message in context: http://wikimedia.7.x6.nabble.com/Failed-jobs-will-never-run-again-tp5049689p... Sent from the Wikipedia Developers mailing list archive at Nabble.com.
Wikitech-l mailing list Wikitech-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikitech-l
wikitech-l@lists.wikimedia.org