Ticket #243 (closed defect: wontfix)

Opened 2 years ago

Last modified 1 year ago

Unable to stop Mulgara - possible thread deadlock

Reported by: rich Assigned to: ronald
Priority: medium Milestone:
Component: topaz-mulgara Version: 0.5-rc1
Keywords: mulgara Cc: rcave@plos.org
Blocking: Blocked By:

Description

"VM Periodic Task Thread" prio=1 tid=0x0000002ac57fe730 nid=0x3434 waiting on condition

Dependency Graph

Attachments

mulgara.out (74.3 kB) - added by rich on 12/18/06 10:07:47.
mulgara.log.2006-12-17.tgz (1.9 MB) - added by rich on 12/19/06 17:29:07.

Change History

12/18/06 10:07:47 changed by rich

  • attachment mulgara.out added.

12/18/06 11:11:55 changed by ronald

What's in the mulgara.log ? It looks like there's problems while shutting down the resolvers, as the RMI thread is still running.

12/19/06 13:04:00 changed by ronald

  • priority changed from unassigned to critical.
  • status changed from new to assigned.
  • version deleted.

12/19/06 17:29:07 changed by rich

  • attachment mulgara.log.2006-12-17.tgz added.

12/19/06 19:16:24 changed by ronald

  • cc set to rcave@plos.org.
  • keywords set to mulgara.
  • version set to 0.5-rc1.
  • priority changed from critical to medium.

From logs it is clear that the resolvers and database were shut down, i.e. the logs and database were flushed and closed properly. All the JVM is waiting for is for the RMI Reaper thread to terminate.

As to why that thread did not terminate in this case, that is unclear. There are two entities that (potentially) use RMI: ehcache and jotm. However, I don't think jotm is actually using RMI. ehcache goes to some trouble to properly shut down RMI, and this issue is discussed here: http://archives.java.sun.com/cgi-bin/wa?A2=ind0608&L=rmi-users&O=D&P=367 The fixes mentioned there are in version 1.2.3, which we are using.

Also, I have been unable to reproduce this, and from the logs it seems that in most cases the shutdown proceeds correctly. Because of this and the fact that the most critical part, namely the proper closing of the logs and database, is happening, I'm lowering the priority of this issue.

08/07/07 16:25:51 changed by

  • milestone deleted.

Milestone Bugs deleted

11/08/07 17:50:57 changed by rich

  • status changed from assigned to closed.
  • resolution set to wontfix.

Closing ticket as it's well past teething years.