[progress Communities] [progress Openedge Abl] Forum Post: Load Average Skyrocket In Deadly...

[Simon L. Prinsloo]

Hi My client experience a problem where two AppServer processes ends up in a deadly embrace situation. As both calls are made from a batch job, this should inherently only lock up processing of the message queue, but its effect goes much wider. The load average of the machine skyrockets, causing system wide response problems for online users. Is it expected that the dead lock situation will cause a race condition on the machine and how can we avoid the race condition? We need to be able to keep the lock in place long enough to identify the offending processes without impacting the interactive users, as the remedial action for the response problem comes down to restarting the batch jobs and their associated AppServer, thereby loosing all evidence of the code causing the deadly embrace before we can pinpoint the source. Regards Simon

Continue reading...