Bug 4230 - DRS deadlocks in thread scheduling
: DRS deadlocks in thread scheduling
: Data Replication
Data Replication Service
: unspecified
: All All
: P3 critical
: ---
Assigned To:
  Show dependency treegraph
Reported: 2006-02-22 17:07 by
Modified: 2006-02-22 17:12 (History)



You need to log in before you can comment on or make changes to this bug.

Description From 2006-02-22 17:07:05
Deadlock is due to non-daemon threads waiting on other non-daemon threads (and
other external resources), thus not relinquishing the worker thread and thus not
returning the worker thread to the work manager's thread pool. Since other
threads cannot be scheduled, the net result is a deadlocked DRS (if not work
manager in total).
------- Comment #1 From 2006-02-22 17:07:50 -------
Fix is to change Discover, Transfer, and TransferRequest {Worker}s to return
true from isDaemon method.
------- Comment #2 From 2006-02-22 17:12:16 -------
Committed to trunk and globsu_4_0_branch