Percona Server 5.6 RC with thread pool crashes under load
Bug #1191375 reported by
Vadim Tkachenko
This bug affects 1 person
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Percona Server moved to https://jira.percona.com/projects/PS |
Fix Released
|
High
|
Unassigned | ||
5.1 |
Invalid
|
Undecided
|
Unassigned | ||
5.5 |
Invalid
|
Undecided
|
Unassigned | ||
5.6 |
Fix Released
|
High
|
Unassigned |
Bug Description
Running PS 5.6 with pool of threads I observe regular crashes.
There is gdb backtrace
consider following scenario create_ thread -> create_worker ->
-> mysql_thread_create -> spawn_thread_v1
* thread t1 is running
* t1 calls wait_begin -> wake_or_
current thread's PFS_thread* remembered here as parent thread's
PFS_thread* for newly created thread t2
* t1 finishes it's wait, sends result to client and finishes itself
* t2 is just about to start, it tries to attach instrumentation to
itself. It wants to access PFS_thread* of t1 which is no longer
available.
this scenario is impossible in Percona Server 5.5 as pointers
to parent thread's structures are not used
I'll try to reproduce it with DEBUG_SYNC or using other tricks