Hi Mark,
Do you have multiple CEs? It could be that only one is reporting in?
Try running this on your APEL box:
mysql -u accounting - p<apel_db_password> accounting < <(echo 'SELECT COUNT(RecordIdentity),SUBSTRING(ExecutingCE FROM 1 for 9),EventDate from LcgRecords WHERE EventDate>='20130401' AND EventDate<='20130619' GROUP BY SUBSTRING(ExecutingCE FROM 1 for 9),EventDate ORDER BY EventDate,SUBSTRING(ExecutingCE FROM 1 for 9);' )
(Note our CEs are uniquely identified by the first 9 chars - heplnx20x - change the SUBSTRING length for your CE names.
That should give you the number of jobs per CE per day in the LcgRecords Table, which is what gets sent to the central DB.
Yours,
Chris.
> -----Original Message-----
> From: Testbed Support for GridPP member institutes [mailto:TB-
> [log in to unmask]] On Behalf Of Mark Slater
> Sent: 19 June 2013 09:21
> To: [log in to unmask]
> Subject: Double Checking APEL records
>
> Hi All,
>
> I seem to be having trouble with the APEL reporting at Bham. Both Torque
> and CREAM (or at least the logs /var/torque/server_logs and
> /var/log/cream/accounting/blahp.log-*) indicate that I'm running
> 3000-5000 jobs per day whereas those reported by APEL seem to be a lot
> lower:
>
> [root@epgr11 ~]# grep "Joined accounting records" /var/log/apel.log | tail -n
> 20 Tue Jun 11 06:41:19 UTC 2013: apel-publisher - Number of Joined
> accounting records: 5019 Tue Jun 11 06:41:19 UTC 2013: apel-publisher -
> Number of Joined accounting records: 0 Wed Jun 12 06:41:09 UTC 2013: apel-
> publisher - Number of Joined accounting records: 422 Wed Jun 12 06:41:09
> UTC 2013: apel-publisher - Number of Joined accounting records: 0 Thu Jun
> 13 06:41:21 UTC 2013: apel-publisher - Number of Joined accounting records:
> 5883 Thu Jun 13 06:41:21 UTC 2013: apel-publisher - Number of Joined
> accounting records: 0 Fri Jun 14 06:41:16 UTC 2013: apel-publisher - Number
> of Joined accounting records: 2666 Fri Jun 14 06:41:16 UTC 2013: apel-
> publisher - Number of Joined accounting records: 0 Sat Jun 15 06:41:13 UTC
> 2013: apel-publisher - Number of Joined accounting records: 427 Sat Jun 15
> 06:41:13 UTC 2013: apel-publisher - Number of Joined accounting records: 0
> Sun Jun 16 06:41:11 UTC 2013: apel-publisher - Number of Joined accounting
> records: 121 Sun Jun 16 06:41:11 UTC 2013: apel-publisher - Number of Joined
> accounting records: 0 Mon Jun 17 06:41:07 UTC 2013: apel-publisher -
> Number of Joined accounting records: 0 Mon Jun 17 06:41:07 UTC 2013: apel-
> publisher - Number of Joined accounting records: 0 Tue Jun 18 06:41:20 UTC
> 2013: apel-publisher - Number of Joined accounting records: 5800 Tue Jun 18
> 06:41:20 UTC 2013: apel-publisher - Number of Joined accounting records: 0
> Wed Jun 19 06:43:53 UTC 2013: apel-publisher - Number of Joined accounting
> records: 3632 Wed Jun 19 06:43:53 UTC 2013: apel-publisher - Number of
> Joined accounting records: 0 Wed Jun 19 07:54:28 UTC 2013: apel-publisher -
> Number of Joined accounting records: 0 Wed Jun 19 07:54:28 UTC 2013: apel-
> publisher - Number of Joined accounting records: 0
>
>
> In fact, just looking at torque jobs completed in the last 9 days
> (47257) it seems this is a lot higher than what the accounting portal is saying
> for the whole of June (38,367)! I don't think I changed anything recently and
> some data does make it to APEL, it just seems about half of what I expect. Is
> there any way to check what my local APEL box thinks the numbers are and
> try to debug the differences?
>
> Any help is very much appreciated!
>
> Thanks,
>
> Mark
--
Scanned by iCritical.
|