I have one question regarding to failover to tuxedo 9.1 from OSB 10.3.1 through WTC on Weblogic.
We faced the issue that if one node from the 2 nodes tuxedo cluster is partitioned the communication from OSB to the Tuxedo partitioned node is going to timeout…
We have setup-ed the WTC failover on Weblogic WTC Import using the remote access point list on WTC using comma separator as failover sign:
Remote Access Point List: TUXOSBA,TUXOSBB
In OSB we have setup-ed the tuxedo service as one endpoint to that WTC failovered service definition. The service timeout on OSB is set to 30 secs.
It seems that the failover is just for network unavailability of the service, but if one tuxedo node is partitioned, the network address is still accessible, but the requests timeouted. So the failover will not be activated (service is ready, but it takes too long)
– after all the Weblogic threads are exhausted and the Weblogic/OSB is dead for any new requests.
How to detect this issue and if timeout/s occur/s the failover on WTC or on OSB WILL be activated??
(we can define more endpoints on business service, but the failover will not be activated too – if timeout will occur)
Is the service you are trying to call available on both machines in the Tuxedo domain? If so, you should still be able to get requests processed even though the domain is partitioned. If you can't run the service on both machines, then you might try using the Tuxedo Event Broker and subscribe to partition events and then execute a command to disconnect the link. That will force WTC to fail over to other machine.
Oracle Tuxedo Chief Architect