[GLASSFISH-15828] [Test Issue] On some GF-HA functional test setups, some metro-ha tests do not complete and get reported as failure. Created: 03/Feb/11  Updated: 07/Feb/11

Status: Open
Project: glassfish
Component/s: sqe-test
Affects Version/s: 3.1_b40
Fix Version/s: None

Type: Bug Priority: Major
Reporter: varunrupela Assignee: varunrupela
Resolution: Unresolved Votes: 0
Labels: None
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified

Attachments: Zip Archive in-order-test-failure.zip    
Issue Links:
blocks GLASSFISH-15645 [UB][RM-HA] [to-be-release-noted] RM ... Resolved
Tags: 3_1-exclude


On some of the GF-HA Functional Test setups, the InOrder Metro-HA tests get Interrupted by TestNG due to a configured 5 minute timeout. This interruption results in a failure

Responses on these test setups are slow due to which the RM-Client ends up sending a AckRequested, which further slows down the actual response. After a few messages, the LB begins to return a 503 Service Unavailable to the RM-Client. This make the RM-Client retry messages further slowing down rates of responses. Eventually TestNG interrupts the test once it crosses 5 minutes.

The test may need to be retried with fin-level logging and the setup need to be re-looked at to figure the reason for the slowness of the responses. We may also need to decide if the 5 min timeout is justified or if we need to slow-down the AckRequested rate.

Attaching client-output, serving-instance logs and the lb error logs for one such failure.

Generated at Fri Aug 26 00:02:38 UTC 2016 using JIRA 6.2.3#6260-sha1:63ef1d6dac3f4f4d7db4c1effd405ba38ccdc558.