You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@curator.apache.org by "Michael Pawliszyn (JIRA)" <ji...@apache.org> on 2017/03/01 15:45:45 UTC
[jira] [Created] (CURATOR-389) Factory and Default ThreadFactory
inconsistencies and poor thread names
Michael Pawliszyn created CURATOR-389:
-----------------------------------------
Summary: Factory and Default ThreadFactory inconsistencies and poor thread names
Key: CURATOR-389
URL: https://issues.apache.org/jira/browse/CURATOR-389
Project: Apache Curator
Issue Type: Bug
Reporter: Michael Pawliszyn
Assignee: Jordan Zimmerman
If you pass in a ThreadFactory to the CuratorFrameworkFactory that ThreadFactory is used for both the Framework and the ConnectionStateManager. If no ThreadFactory is passed in two thread factories are created so the threads are named based on what class is using the thread, for example:
"Curator-ConnectionStateManager-0" #62 daemon prio=5 os_prio=0 tid=0x00007f3d1272d800 nid=0x5466c waiting on condition [0x00007f3c46620000]
"Curator-Framework-0" #418 daemon prio=5 os_prio=0 tid=0x00007f3d12d29800 nid=0x54ece waiting on condition [0x00007f3ade6ef000]
If there is more than one curator there is no indication on what Zookeeper connections they are curating by looking at the thread dumps. Overall better default thread naming that uses information from the EnsembleProvider would help investigations.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)