Dear All,
As you see, the error comes from 'Publishing resutls to SAME'. SAME
is the new monitoring framework that is in development. Results are already being
published for that one, together with being published to the old database in
production. These messages are harmless. Sorry for not pointing this
out, the error message will be made more verbose to tell that it's harmless.
Judit
On v, már 12, Kyriakos G. Ginis wrote:
> Hi all,
>
> I noticed during the last hours that the SFTs have trouble publishing their
> output on many sites including ours, for example in all:
>
>
> https://lcg-sft.cern.ch/sft/info/ce.phy.bg.ac.yu/sft_2006-03-12_16.05.01.html#sft-job_2006-03-12_18:02:02
> https://lcg-sft.cern.ch/sft/info/heplnx201.pp.rl.ac.uk/sft_2006-03-12_16.05.01.html#sft-job_2006-03-12_17:01:10
> https://lcg-sft.cern.ch/sft/info/ce001.grid.bas.bg/sft_2006-03-12_16.05.01.html#sft-job_2006-03-12_18:01:54
>
>
> there are messages like the following:
>
> Publishing results to SAME... result-tuples-sft-apel.txt: Publishing failed: Couldn't contact publisher webservice!
> DONE
> result-tuples-sft-brokerinfo.txt: Publishing failed: Couldn't contact publisher webservice!
> DONE
> result-tuples-sft-caver.txt: Publishing failed: Couldn't contact publisher webservice!
> DONE
>
>
> and earlier in
>
> https://lcg-sft.cern.ch/sft/info/ce01.isabella.grnet.gr/sft_2006-03-12_01.05.01.html#sft-job_2006-03-12_01:32:45
>
> Publishing results to SAME... result-tuples-sft-apel.txt: : {'hostname':
> 'grvw003.cern.ch'}>
> Publishing failed: java.lang.Exception: Unable to Insert data : Caused
> by : ORA-01000: maximum open cursors exceeded
>
> DONE
> result-tuples-sft-brokerinfo.txt: : {'hostname': 'grvw003.cern.ch'}>
> Publishing failed: java.lang.Exception: Unable to Insert data : Caused
> by : ORA-01000: maximum open cursors exceeded
>
> DONE
> result-tuples-sft-caver.txt: : {'hostname': 'grvw003.cern.ch'}>
> Publishing failed: java.lang.Exception: Unable to Insert data : Caused
> by : ORA-01000: maximum open cursors exceeded
>
>
> Sometimes it takes so long for the publishing script to timeout, that
> the batch system kills the job, especially if it is running on a 'short'
> queue with max wallclocktime around 1 hour. As a result the real problem
> is masked, since it appears on the SFT page as a 'Job Submission'
> failure....
>
> Cheers,
>
> --
> Kyriakos Ginis
> HG-01-GRNET team
|