Deadlock in location_monitors update via Hibernate
Description
On a customer system running the latest 1.13 code, we are encountering an issue where the database connection pool (Atomikos) stops successfully allocating new connections. In the poller log, there was a PostgreSQL deadlock detection exception being caught. This may be the root cause of the database connection pool problems and should be fixed.
The error appears to be occurring inside the DefaultPollerBackEnd.checkForDisconnectedMonitors() method.
Acceptance / Success Criteria
None
Lucidchart Diagrams
Activity
Show:
Seth Leger February 10, 2014 at 5:36 PM
I've added locking support to the Criteria API so that you can specify a lock mode for Criteria queries and I changed this particular call to fetch all records in a single database call (instead of fetching everything and iterating over the list in Java like it was doing). This should make the call faster and should prevent deadlocks like this from occurring. Marking as fixed.
On a customer system running the latest 1.13 code, we are encountering an issue where the database connection pool (Atomikos) stops successfully allocating new connections. In the poller log, there was a PostgreSQL deadlock detection exception being caught. This may be the root cause of the database connection pool problems and should be fixed.
The error appears to be occurring inside the DefaultPollerBackEnd.checkForDisconnectedMonitors() method.