[MXS-164] setting log_debug=1 triggers signal 11 crash Created: 2015-05-20 Updated: 2015-11-25 Resolved: 2015-11-25 |
|
| Status: | Closed |
| Project: | MariaDB MaxScale |
| Component/s: | Core |
| Affects Version/s: | 1.1.0 |
| Fix Version/s: | 1.3.0 |
| Type: | Bug | Priority: | Minor |
| Reporter: | Glenn Plas | Assignee: | Massimiliano Pinto (Inactive) |
| Resolution: | Cannot Reproduce | Votes: | 0 |
| Labels: | logging | ||
| Environment: |
DISTRIB_ID=Ubuntu |
||
| Attachments: |
|
||||||||
| Issue Links: |
|
||||||||
| Description |
|
Without any log_debug settings, maxscale actually seems to run fine. Once you set this flag to 1, it will not start and crash with signal 11. It happens, nomatter what config you use, below is a tracer of an unfinished setup but I have a semi-production setup that does exactly the same. All ubuntu 14.04 May 20 11:55:48 ttextlb1 MaxScale[18055]: Home directory : /usr/local/mariadb-maxscale |
| Comments |
| Comment by Massimiliano Pinto (Inactive) [ 2015-06-11 ] |
|
gplv2 In the attached maxScale.cnf I see: #log_debug=1 Is the issue about starting maxscale with that option set or changing the value via maxadmin ? Did you try with MaxScale 1.1.1 as well? Could you please start MaxScale whith gdb ? #gdb ./maxscale and if the issue is found, please type in gdb (gdb) where This way we get full stack trace Thanks |
| Comment by Glenn Plas [ 2015-06-11 ] |
|
"Once you set this flag to 1, it will not start" -> maybe I wasn't too clear on this. it's commented out in the config due the the sequence of troubleshooting these events, but it's a startup issue for me. It will launch but immediately crash and syslog this info. Will test 1.1.1 and get back to you. Not sure if I can do the gdb anytime soon and trigger it. It takes days between crashes and we kind of need this to be up. I will try though. |
| Comment by Glenn Plas [ 2015-06-12 ] |
|
Hi, I've tested switching on trace and debug log using maxadmin. This does not cause a crash so far. So once it is running it seems to be working. I got you the stack trace as this was reproducable. root@ttextlb2:/etc/maxscale/etc# gdb /usr/local/mariadb-maxscale/bin/maxscale [New Thread 0x7ffff1403700 (LWP 15462)] Program received signal SIGSEGV, Segmentation fault. |
| Comment by Glenn Plas [ 2015-06-12 ] |
|
to avoid confusion with paths in the above , I have softlinked the maxscale directory under /etc lrwxrwxrwx 1 root root 27 May 11 00:37 maxscale -> /usr/local/mariadb-maxscale |
| Comment by Dipti Joshi (Inactive) [ 2015-06-16 ] |
|
gplv2 Can you clarify whether you reproduced this with 1.1.1 version of MaxScale ? |
| Comment by Dipti Joshi (Inactive) [ 2015-07-17 ] |
|
gplv2 1.2.0 is out, can you try to reproduce this with 1.2.0 now ? |
| Comment by markus makela [ 2015-11-25 ] |
|
We have not been able to reproduce this so we are closing this for now. If the problem still appears with 1.3.0, this bug can be reopened. |