Hi All,
unfortunately we had a bug in the auto resume functionality, both in v2.5.1 and in the master branch (2.6).
I fixed this and uploaded the fix to github.
Basically there are two possibilities to eliminate the bug:
1. Sync with github (best option, I think)
2. Fix it manually
In case someone prefers the latter option, the diff output is:
@@ -92,14 +92,6 @@ public class TriggerThread extends InternalSession
long rts = resumeTs.longValue();
if (rts <= now) {
sme.resume(sysEnv);
- SDMSAuditTrailTable.table.create(sysEnv, cEnv.uid(), new Long ((new java.util.Date()).getTime()),
- new Integer(SDMSAuditTrail.RESUME),
- new Integer(SDMSAuditTrail.JOB),
- sme.getId(sysEnv), sme.getId(sysEnv),
- Boolean.FALSE,
- null,
- "Automatic resume");
-
} else {
if (rts < nextTime) nextTime = rts;
}
The only thing that has to be done is to eliminate the attempt to write the audit table.
After either option, do the following:
1. Shut down the scheduling server
2. Shut down all jobservers local to the scheduling server (they share the same jar file which has to be replaced)
3. Make the new jar file (make new always works, but a plain make should work too)
4. Put the new jar file to $BICSUITEHOME/lib (if it's not already there)
5. Start the scheduling server
6. Start the jobservers
This procedure will require only a few minutes. And apart from the fact that during that time old jobs won't be reported as finished and no new jobs will be started, it shouldn't have side effects.
(If the development environment isn't part of the server installation, it'll be even possible to create the jar file first. The shutdown and restart of the components will be the only downtime then).
I'm sorry for the hassle.
Best regards,
Ronald