SGAS silent "errors"

[kmjonsson]

Mail from: Andreas Bach [email protected]

Hi,

don't know where the best place to report errors is, so I'll just try
with you, and let you pass it on to whoever gets the SGAS Software
Developer position (:

One of our systems reset its job ids earlier. Without noticing Sigma,
which manages the SGAS server. The SGAS server reports A-OK and such and
such number of jobs registered[1], but (almost) no jobs are there in the
database. No errors in sgas.log, nor postgres-logs.

Turns out the jobs are silently dropped when there is a duplicate
record_id. This behaviour should probably be looked at.

[1] 2013-08-06 01:00:33+0200 [sgas.Authorizer] Authz check: Subject
/O=Grid/O=NorduGrid/CN=host/vilje.hpc.ntnu.no, Action jobinsert,
Context: [('machine_name', 'vilje.hpc.ntnu.no')]. Access allowed: True
2013-08-06 01:00:33+0200 [sgas.PostgreSQLDatabase] Database: 100 job
usage records inserted

Regards,

-- Andreas Bach UNINETT Sigma