orphaned thd with lp:1208493 regression test
Affects | Status | Importance | Assigned to | Milestone | ||
---|---|---|---|---|---|---|
MySQL patches by Codership | Status tracked in 5.6 | |||||
5.5 |
Incomplete
|
Undecided
|
Unassigned | |||
5.6 |
New
|
Undecided
|
Unassigned | |||
Percona XtraDB Cluster moved to https://jira.percona.com/projects/PXC | Status tracked in 5.6 | |||||
5.5 |
Incomplete
|
Undecided
|
Unassigned | |||
5.6 |
Incomplete
|
Undecided
|
Unassigned |
Bug Description
Regression test for lp:1208493 got stuck in:
starting node0, node1...
140408 04:29:00.315 Job 'start_cmd' on 'local1' complete in 7 seconds,
140408 04:29:07.569 Job 'start_cmd' on 'local2' complete in 7 seconds,
wsrep_provider = /tmp/galera/
wsrep_cluster_
starting variable setters...
dropping and joining node...
Dropper Done
Waiting load to end (28700)
Running 'netstat -nat' revealed that there was ongoing connection to mysqld instance:
tcp 0 0 10.0.2.15:3311 10.0.2.15:56480 ESTABLISHED
Also mysql processlist listed the connection:
| 251 | unauthenticated user | connecting host | NULL | Connect | NULL | login | NULL |
Corresponding thd was fround from global_thread_list:
(gdb) p ((THD*)
$19 = 251
(gdb) p ((THD*)
$20 = 0xef2a72 "login"
However, gdb thread listing didn't show any thread that would have been handling the connection and the main thread was already polling listener socket.
So far seen only with 5.6.
description: | updated |
Can you re-try it with latest 5.6 and see if the issue still persist.