[
https://jira.jboss.org/jira/browse/JBPM-2691?page=com.atlassian.jira.plug...
]
Alejandro Guizar commented on JBPM-2691:
----------------------------------------
This update caused regression in SQL Server at JBPM983Test. Sadly, concurrent updates
under snapshot isolation are reported as SQLGrammarException instead of
LockAcquisitionExceptions. Jobs hit by locking exceptions are retried immediately, whereas
jobs hit by other exceptions are not retried even once due to an oversight. Field
Job.retries is initialized to 1, which should mean "retry once". However, if an
exception occurs, the job executor decreases the job's retries by one. With retries
set to 0, query JobSession.getFirstAcquirableJob never selects that job again.
The usefulness of field Job.retries is questionable because there is no way to configure a
higher value other than updating the job itself.
Save exceptions thrown while executing a job in a separate
transaction
----------------------------------------------------------------------
Key: JBPM-2691
URL:
https://jira.jboss.org/jira/browse/JBPM-2691
Project: jBPM
Issue Type: Bug
Affects Versions: jBPM 3.2.8
Reporter: Alejandro Guizar
Assignee: Alejandro Guizar
Fix For: jBPM 3.2.9
Attachments: JobExecutorThread.diff
Original Estimate: 0 minutes
Remaining Estimate: 0 minutes
Currently the stack trace printout is saved in the same transaction provided the
exception is not related to persistence. This scheme has two drawbacks:
1. Exceptions not related to persistence may still leave the job and the process instance
as a whole in an inconsistent state.
2. Persistence exceptions are never saved.
Hence saving the stack trace in a separate transaction would be preferable.
--
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators:
https://jira.jboss.org/jira/secure/Administrators.jspa
-
For more information on JIRA, see:
http://www.atlassian.com/software/jira