Hello,
Following instructions provided at
http://www.globus.org/toolkit/docs/4.0/execution/wsgram/WS_GRAM_Audit_Logging.html
we have activated WS GRAM audit logging to a PostgreSQL database.
Unfortunately the connection from GRAM to the audit database is lost,
whenever the database is restarted. Thereafter no audit information
about new jobs is written to the database. A way to restore the
connection is to restart of the Globus container. However, this is no
practical solution if many Globus nodes share the same audit database.
If the database goes down, all Globus containers would have to be
restarted. This of course complicated due to permissions issues and a
missing automatic restart mechanism of the Globus containers in those
situations.
In consideration of this problem we have also tried to use the Pre WS
GRAM audit logging discussed at
http://www.globus.org/toolkit/docs/4.0/execution/prewsgram/Pre_WS_GRAM_Audit_Logging.html
.
Despite following all steps until Section 7.3 no audit files were found
in the designated folder (/tmp/globus/).
Any help with getting audit logging to work would be very appreciated.
Frank
