[MXS-1585] Crash in MaxScale 2.1.12 Created: 2017-12-19  Updated: 2020-08-25  Resolved: 2017-12-29

Status: Closed
Project: MariaDB MaxScale
Component/s: readwritesplit
Affects Version/s: 2.1.12
Fix Version/s: 2.1.13

Type: Bug Priority: Blocker
Reporter: Chris Calender (Inactive) Assignee: markus makela
Resolution: Fixed Votes: 0
Labels: None


 Description   

2017-12-19 16:28:26 error : (341578) [MySQLClient] Routing the query failed. Session will be closed.
2017-12-19 16:42:00 alert : (341585) Fatal: MaxScale 2.1.12 received fatal signal 11. Attempting backtrace.
2017-12-19 16:42:00 alert : (341585) Commit ID: source-build System name: Linux Release string: undefined
2017-12-19 16:42:00 alert : (341585) /usr/bin/maxscale() [0x405dd7]
2017-12-19 16:42:00 alert : (341585) /lib/x86_64-linux-gnu/libpthread.so.0(+0xf890) [0x7f175cd8d890]
2017-12-19 16:42:00 alert : (341585) /usr/lib/x86_64-linux-gnu/maxscale/libreadwritesplit.so(+0x5f11) [0x7f1757980f11]
2017-12-19 16:42:00 alert : (341585) /usr/lib/x86_64-linux-gnu/maxscale/libreadwritesplit.so(handle_master_is_target+0x73) [0x7f1757982123]
2017-12-19 16:42:00 alert : (341585) /usr/lib/x86_64-linux-gnu/maxscale/libreadwritesplit.so(route_single_stmt+0xbc) [0x7f175798255c]
2017-12-19 16:42:00 alert : (341585) /usr/lib/x86_64-linux-gnu/maxscale/libreadwritesplit.so(+0x37a5) [0x7f175797e7a5]
2017-12-19 16:42:00 alert : (341585) /usr/lib/x86_64-linux-gnu/maxscale/libMySQLClient.so(+0x2a9e) [0x7f1756886a9e]
2017-12-19 16:42:00 alert : (341585) /usr/lib/x86_64-linux-gnu/maxscale/libmaxscale-common.so.1.0.0(+0x46b18) [0x7f175d4b6b18]
2017-12-19 16:42:00 alert : (341585) /usr/lib/x86_64-linux-gnu/maxscale/libmaxscale-common.so.1.0.0(poll_waitevents+0x3be) [0x7f175d4b797e]
2017-12-19 16:42:00 alert : (341585) /usr/bin/maxscale(_Z18worker_thread_mainPv+0x99) [0x406269]
2017-12-19 16:42:00 alert : (341585) /lib/x86_64-linux-gnu/libpthread.so.0(+0x8064) [0x7f175cd86064]
2017-12-19 16:42:00 alert : (341585) /lib/x86_64-linux-gnu/libc.so.6(clone+0x6d) [0x7f175b63f62d]



 Comments   
Comment by markus makela [ 2017-12-29 ]

This appears to be a bug in the code that handles queries that are routed to the master. The detection of a lost master was done incorrectly.

Comment by markus makela [ 2017-12-29 ]

Using master_failure_mode=fail_instantly should prevent the crash but it will prevent read-only usage.

Generated at Thu Feb 08 04:07:53 UTC 2024 using Jira 8.20.16#820016-sha1:9d11dbea5f4be3d4cc21f03a88dd11d8c8687422.