In our production server we faced this typical issue and initially thought of having problem in our JCA adapter. The issue was few BPEL process faulted instance starts automatically and again faulted for same reason and we had no control over its start time. As shown in below snap shot our BPEL composite first faulted on 29th August 7:33:59 PM and again starts on 30th August 00:00:36 AM. As a result huge number of faulted instance piled up and crashes the server.
As a fix we shift our thoughts from JCA adapter Retrying bug to SOA Server Recovery settings. Initially we are searching for bug in JCA adapter, as we thought that JCA adapter retry settings make that happens. But actually SOA server automatic Recovery Settings tried that faulted instances to recover for a certain period of time. Please refer below screen shot for default values of SOA Server Automatic Recovery.
Detail description for those attributes are explained by oracle, please refer http://download.oracle.com/docs/cd/E14571_01/relnotes.1111/e10133/soa.htm#RNLIN1052.
Step 1: As a fix we went to our Production server and expand SOA and then right click to soa-infra and select BPEL properties. Please refer below screen shot.
Step 2: Select “More BPEL Configuration Properties” (Refer Below Screen shot)
Step 3: Select “RecoveryConfig” (Refer Below Screen shot)
Step 4: Change Stop Window Time to 00:00 (same as Start Window time , so Recovery Period becomes negligible).
No comments:
Post a Comment