Re: [javagroups-users] Concurrent startup of many channels doesn't stabilize
Brought to you by:
belaban
From: Bela B. <be...@ya...> - 2008-06-17 07:18:31
|
Vladimir Blagojevic wrote: > Robert Newson wrote: >> Ok, 50 runs now. This is better. I'm going to leave it running until >> the morning, then I'll tweak the test to use the shared transport >> (adding singleton_name="something" to the config) and start over. >> >> > Test can fail due to timeout i.e members do not merge in given 120 sec > timeout and the verification code checks that each channel has x many > members after that timeout. I really have a hard time believing that the > lockup can occur as it did before the latest fixes. If it does then > please take a stacktrace! I have increased timeout to 160 seconds. Ah, that's probably what Robert alluded to when he mentioned JUnit failures: assertions which were not true. Should we decrease the merge min (5000) and max (10000) timeouts even further ? -- Bela Ban Lead JGroups / JBoss Clustering team JBoss - a division of Red Hat |