a similar thing happend to some our users when some their jobs remained
stuck in condor with an expired proxy.
To solve this problem:
1) su - edguser
2) check how many jobs are in a held state in condor:
3) check each of these held jobs to see the user they belong
/opt/condor/bin/condor_q -long <condor id>
4) then remove them:
/opt/condor/bin/condor_rm <condor id> ... <condor id>
Guillaume PHILIPPON ha scritto:
> Some of my users complained me that they can't launch any job, RB told
> them that their proxy is expired even if voms-proxy-info -all is ok.
> It seems happen when they launch a job (with a valid proxy) which not
> finish before proxy expiration.
> After that, they can't launch any job with a valid proxy. RB seems not
> check if a new proxy have been created.
> A stop/start of edg-* service in our RB resolve the problem.
> Did somebody already see this problem?
> We are on Glite update 20 and the last change of lcg-rb package was on
> update 17 (we use it since it was release).
> Guillaume PHILIPPON Mail : [log in to unmask]
> LAL (UMR8607) / CNRS Tel : +33 1 64 46 82 71
Dr. Alessandro Paolini
INFN - CNAF
Viale Berti Pichat 6/2
tel: +39 051 6092723
fax: +39 051 6092746
"credo nel potere del riso e delle lacrime"
"come antidoto all'odio ed al terrore"
"un giorno senza un sorriso"
"è un giorno perso" >>> Charlie Chaplin