Details
-
Sub-task
-
Status: Closed
-
Major
-
Resolution: Later
-
None
-
None
-
None
Description
Looking at thread counts reported by the nice 'after:' log message, I see the likes of TestReplicationKillSlaveRS and all of its derivatives reporting hundreds of threads (shows as over 1000 in one run here).
Running the test standalone, I see the above run up the threads to 700 (only!). Need to figure why the discrepancy.
The attached patch makes some difference – cutting down the client thread count – but not enough. Let me look more.