[jira] [Commented] (HBASE-12937) Infinite tracing loop brings down region server

2015-01-28 Thread stack (JIRA)

[ 
https://issues.apache.org/jira/browse/HBASE-12937?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14295449#comment-14295449
 ] 

stack commented on HBASE-12937:
---

[~giacomotaylor] HTrace is guest of the hbase process. If it expects the host 
to do resets on it on a period, it is a bad 'guest'. Lets try fix over in 
HTRACE w/o having to have hbase or hdfs or phoenix or whoever have to reach 
into HTRACE implementation detail to do null-sets.

 Infinite tracing loop brings down region server
 ---

 Key: HBASE-12937
 URL: https://issues.apache.org/jira/browse/HBASE-12937
 Project: HBase
  Issue Type: Bug
Reporter: James Taylor

 See HTRACE-92 for more detail. I'm hoping that adding calls to 
 Tracer.getInstance().continueSpan(null) in the correct place(s) will do the 
 trick. This is rendering tracing unusable by Phoenix. Not sure if it impact 
 general HTrace use in HBase or not.
 FYI, [~samarthjain], [~jesse_yates].



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)


[jira] [Commented] (HBASE-12937) Infinite tracing loop brings down region server

2015-01-28 Thread James Taylor (JIRA)

[ 
https://issues.apache.org/jira/browse/HBASE-12937?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14295417#comment-14295417
 ] 

James Taylor commented on HBASE-12937:
--

My thinking was that HTrace is not going to know when to reset the thread local 
state through Tracer.getInstance().continueSpan(null), while a client (HBase in 
this case) would. Is that not the case, [~stack]?

 Infinite tracing loop brings down region server
 ---

 Key: HBASE-12937
 URL: https://issues.apache.org/jira/browse/HBASE-12937
 Project: HBase
  Issue Type: Bug
Reporter: James Taylor

 See HTRACE-92 for more detail. I'm hoping that adding calls to 
 Tracer.getInstance().continueSpan(null) in the correct place(s) will do the 
 trick. This is rendering tracing unusable by Phoenix. Not sure if it impact 
 general HTrace use in HBase or not.
 FYI, [~samarthjain], [~jesse_yates].



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)