Monitor DB Locks that cause a lock wait timeout

[eddiej]

We are running Syteline 5.00.3 with Progresss 9.1c. Our configuration is multi-db/multi-site.

We occasionally have users (and sometimes processes) that cause record locks with some of our mission critical apps for an extended period of time. Currently we are configured such that if a record lock takes place with one of our mission critical apps, and the lock lasts longer then an hour, the app dies, and the message "Lock wait timeout.." is displayed. Unfortunately when this happens we have no record of who or what caused the record lock. All we can do is restart the process and see if it happens again. Short of having someone use promon to look for record locks all day, is there some way we can be alerted when record locks last over a certain amount of time?

 

[taqvia]

Look at the stopinfo file generated which will give a brief idea where the locking contentions were when the there was abnormal abort of the transcation. Also a script can be written to trace the flow or session debug mode can be set to true which will generate a text file for a particular session and store it in the temp directory specified(if locking contentions are being encountered for ceratin clients) remember to unset the session debug option as it generates huge files.

 

[TomBascom]

ProTop provides a realtime view of record locks and active transactions. Both displays will give you good insight into what's going on and who is responsible.