[MDEV-16173] Galera cluster 10.2.14 crash after ALTER table Created: 2018-05-15  Updated: 2018-07-02  Resolved: 2018-07-02

Status: Closed
Project: MariaDB Server
Component/s: Galera
Affects Version/s: 10.2.14
Fix Version/s: N/A

Type: Bug Priority: Blocker
Reporter: Cristian Pirtea Assignee: Unassigned
Resolution: Incomplete Votes: 1
Labels: crash, galera, innodb, need_feedback
Environment:

centos 7.4



 Description   

I have three hosts running MariaDB with Galera Cluster.
Every worked fine until we have to alter index on one of our tables:

ALTER TABLE  'db_name`.`table_name` CHANGE `send_by` `send_by` INT(30) NULL DEFAULT NULL;

Before the column `send_by` was of type "VARCHAR" , it has an index on it also.
All galera nodes(3) crashed with signal 11 after that query:

May 14 21:06:20 mysqld: 180514 21:06:20 [ERROR] mysqld got signal 11 ;
May 14 21:06:20 mysqld: This could be because you hit a bug. It is also possible that this binary
May 14 21:06:20 mysqld: or one of the libraries it was linked against is corrupt, improperly built,
May 14 21:06:20 mysqld: or misconfigured. This error can also be caused by malfunctioning hardware.
May 14 21:06:20 mysqld: To report this bug, see https://mariadb.com/kb/en/reporting-bugs
May 14 21:06:20 mysqld: We will try our best to scrape up some info that will hopefully help
May 14 21:06:20 mysqld: diagnose the problem, but since we have already crashed,
May 14 21:06:20 mysqld: something is definitely wrong and this may fail.
May 14 21:06:20 mysqld: Server version: 10.2.14-MariaDB-log
May 14 21:06:20 mysqld: key_buffer_size=134217728
May 14 21:06:20 mysqld: read_buffer_size=4194304
May 14 21:06:20 mysqld: max_used_connections=1002
May 14 21:06:20 mysqld: max_threads=2002
May 14 21:06:20 mysqld: thread_count=830
May 14 21:06:20 mysqld: It is possible that mysqld could use up to
May 14 21:06:20 mysqld: key_buffer_size + (read_buffer_size + sort_buffer_size)*max_threads = 16574180 K  bytes of memory
May 14 21:06:20 mysqld: Hope that's ok; if not, decrease some variables in the equation.
May 14 21:06:20 mysqld: Thread pointer: 0x7f44a40009a8
May 14 21:06:20 mysqld: Attempting backtrace. You can use the following information to find out
May 14 21:06:20 mysqld: where mysqld died. If you see no messages after this, something went
May 14 21:06:20 mysqld: terribly wrong...
May 14 21:06:20 kernel: mysqld[31633]: segfault at 2 ip 00007f5553a4d118 sp 00007f48af74b1c0 error 4 in libgcc_s-4.8.5-20150702.so.1[7f5553a3e000+15000]

As I saw there was another issue regarding galera cluster crush based on alter unique index
MDEV-9416
Please suggest how to fix this.



 Comments   
Comment by Elena Stepanova [ 2018-06-01 ]

The issue that you mentioned was fixed 1.5 years ago.
For yours, there isn't much to work with here.
Does it happen every time when you attempt this ALTER, or was it a one-time thing?
Did you get a coredump from the crash?
If not and if the problem is repeatable, can you produce a coredump and get a proper stack trace?
Please also provide your cnf file(s) and the definition of the table which you're trying to ALTER (the output of SHOW CREATE TABLE table_name; SHOW INDEX IN table_name;)

Generated at Thu Feb 08 08:26:55 UTC 2024 using Jira 8.20.16#820016-sha1:9d11dbea5f4be3d4cc21f03a88dd11d8c8687422.