Not clearing the job queue

Hello everyone,


We've been experiencing some oddities in our database and we hope to find some help around here. The scenario is as follows:

Our job queue shows 13 pages of jobs, updated in the last 5 days, in which we find lots of jobs - actually most of them from 3 days ago - with the progress showing "transcoding clip" and the status as "RUN".
When we try to cancel any of those jobs we get an error message that says: "Error cancelling - this job is currently not running".

We have tried flushing the event queue from Terminal using the following commands:

fcsvr_client flush_response_queue

fcsvr_run psql px pxdb -c "vacuum analyze verbose;"

but none of them solved our problem. We cannot clear the event tables in the db. We have even tried a more desperate solution by accessing the db via Navicat and manually deleting each and every row (!!!), visually clearing the tables.
BUT on the next day they did not only reappear but double in quantity. So now we have 18,826 pages in our table which seems to grow indefinitely (and for no apparent reason).

We fear that our problem may get worse and cause us bigger problems. Any help would be very appreciated.

Thank you in advance,
Bruno

Stop screwing around with Apple's 'friendly' little DB utilities and just flush them all out manually using the PostgreSQL backend.

fcsvr_run psql px pxdb -c 'delete from pxjob;'
fcsvr_run psql px pxdb -c 'delete from pxeventresponse;'
fcsvr_run psql px pxdb -c 'delete from pxevent;'

That'll kill all the jobs in the table and take out anything that's jammed in the event tables. Now if they fill up again - then you have something like a watcher or a scan that's re-kicking them off?

Dear John,


as I've said before, the pxevent and pxeventresponse issues are dealed with BUT we are running into the same problem posted in my original post:
when you perform a "search all jobs" in FCS, you get a lot of jobs with the status set to RUN eventhough they are not running. If we try to CANCEL it, it won't and gives us an error message saying that this job is currently not running.
When you "get info" it says the job failed to run twice which should automatically turn its status to FAIL.


Have you seen this before and know what is causing this?
I'm attaching some images so you can get the picture (just in case I was not clear enough).

Search all jobs queue: searchalljobs.jpg

Error trying to cancel: errorcanceling.jpg

Get info: get_info.jpg

Thanks