• Bug
  • Status: Closed
  • 2 Major
  • Resolution: Fixed
  • drb
  • Reporter: tgautier
  • September 29, 2008
  • 0
  • Watchers: 0
  • April 10, 2009
  • November 13, 2008

Attachments

Description

I am running the test provided here:

http://forums.terracotta.org/forums/posts/list/1425.page

When I Ctrl-C a client, I get this error in the others:

Wake up, but local generation the same as new generation. Entering debug block. Current Status CyclicBarrier: – parties: 5 – barrierCommand: null – generation: java.util.concurrent.CyclicBarrier$Generation@3bf636 – trip: com.tcclient.util.concurrent.locks.ConditionObject@a1602b Is local generation equal to new generation: true Current Status CyclicBarrier: – parties: 5 – barrierCommand: null – generation: java.util.concurrent.CyclicBarrier$Generation@3bf636 – trip: com.tcclient.util.concurrent.locks.ConditionObject@a1602b

Comments

Taylor Gautier 2008-10-20

looks like a dupe.

Chris Dennis 2008-10-21

Reordered mutations to force barrier tripping to occur in one transaction. A dying JVM cannot then expose the barrier in a mutilated state.

Fiona OShea 2008-11-07

Please merge trunk:10630 to 2.7 branch..thanks

Fiona OShea 2008-11-07

Reopening to make sure that we merge to 2.7…

Chris Dennis 2008-11-07

Merged changes into 2.7 branch (r10771). Reassigning to IRB (should we include the CyclicBarrier debug adapter in transformation?)

Chris Dennis 2008-11-07

Should we include the debug adapter in 2.7 branch. This may still fail (see CDV-954) and without the debug printout it will appear like very odd behavior.

Fiona OShea 2008-11-13

Per Steve, lets leave it in trunk only.