[MXS-269] Crash in MySQL backend protocol Created: 2015-07-15  Updated: 2015-08-21  Resolved: 2015-08-21

Status: Closed
Project: MariaDB MaxScale
Component/s: mariadbbackend
Affects Version/s: 1.2.1
Fix Version/s: 1.3.0

Type: Bug Priority: Blocker
Reporter: martin brampton (Inactive) Assignee: martin brampton (Inactive)
Resolution: Fixed Votes: 0
Labels: None
Environment:

CentOS 6.5



 Description   

During tests with mysqlslap running against the current develop branch, MaxScale crashed with a segment fault:
Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 0x7fffc8dfa700 (LWP 11944)]
0x00007fffc81e25ae in gw_read_backend_event (dcb=0x7fffbc00c000) at /usr/local/skysql/maxscale/MaxScale/server/modules/protocol/mysql_backend.c:455
455 CHK_SESSION(session);

This suggests that the session pointed to by the DCB passed to gw_read_backend_event was freed during processing. Unfortunately, I was looking for other things and did not gather any additional diagnostic information. The problem does not readily repeat.



 Comments   
Comment by Dipti Joshi (Inactive) [ 2015-07-15 ]

martin brampton: Did you use debug build ? Do you have core dump ?

Comment by martin brampton (Inactive) [ 2015-07-17 ]

I believe that this problem is related to persistent connections.

Comment by Johan Wikman [ 2015-08-21 ]

Martin, any news on this one?

Comment by martin brampton (Inactive) [ 2015-08-21 ]

Yes, I believe it was fixed as part of the persistent connections development. It related to how to safely detach the connection from the old session and make it available for attachment to a new one. I think it was only loosely related to other crashes.

Comment by martin brampton (Inactive) [ 2015-08-21 ]

No recurrence of problem despite long test runs

Generated at Thu Feb 08 03:58:03 UTC 2024 using Jira 8.20.16#820016-sha1:9d11dbea5f4be3d4cc21f03a88dd11d8c8687422.