Bugzilla – Bug 4229
Improve SEG fatal Error logging
Last modified: 2006-02-28 13:59:25
You need to log in before you can comment on or make changes to this bug.
currently the only errors that show up in the GT container log file are those from the dynamic loader when it starts the seg. Other fatal errors should be caught and logged. For example, if the SEG does not have permissions to read the local Resource Manager's log file.
I've updated all of the SEG modules we provide in GT4 (in the GT4 branch). These modules will now have a default c debug logging configuration with ERROR level messages sent to stderr in the SEG_PBS_DEBUG (or other scheduler) environment vars are not set. Here are some examples of common SEG problems and how they will appear: Fork: setup package not run: [ERROR] Error retrieving log_path attribute from $GLOBUS_LOCATION/etc/globus-fork.conf log file not readable by SEG: [ERROR] Fatal error (open /home/bester/development/globus_4_0_branch/INSTALL/var/globus-fork.log): Permission denied log file not present: [ERROR] Error reading logfile /home/bester/development/globus_4_0_branch/INSTALL/var/globus-fork.log: No such file or directory After the Java code is done with it and it is filtered through log4j, they end up looking like this: 2006-02-28 14:52:21,623 ERROR monitoring.SchedulerEventGenerator [Thread-2,run:193] SEG Terminated with [ERROR] Error reading logfile /home/bester/development/globus_4_0_branch/INSTALL/var/globus-fork.log: No such file or directory
