Why does galera crash by a simple ALTER -statement - mariadb

Question

I have a Mariadb 10.2.14 5-nodes Galera server. Simple straightforward database almost 20G. No triggers. A lot of INDEXes and FOREIGN KEYS.
Where I try to ALTER a empty or small table (add a field) via the command line MySQL on one of the multi-masters then the whole cluster crashes, why? I never have had this problem on other Galera Systems. RedHat 6.10 is the OS.
Can someone help?
This is the error-log on one of the servers:
While update a simple table with an simple alter statement the 5-node multimaster Galera stops working an the table gets corrupted. This has happened several times nog with different tables and simple alterstatements (without triggers).
The mysql-errorlog shows this:
2019-01-15 10:47:19 140487941920512 [Note] WSREP: Member 1.0 (server.company.local) synced with group.
2019-01-15 11:07:45 140487941920512 [Note] WSREP: Member 1.0 (server.company.local) desyncs itself from group
2019-01-15 11:07:46 140487941920512 [Note] WSREP: Member 1.0 (server.company.local) resyncs itself to group
2019-01-15 11:07:46 140487941920512 [Note] WSREP: Member 1.0 (server.company.local) synced with group.
2019-01-15 11:27:40 140487941920512 [Note] WSREP: Member 1.0 (server.company.local) desyncs itself from group
2019-01-15 11:27:41 140487941920512 [Note] WSREP: Member 1.0 (server.company.local) resyncs itself to group
2019-01-15 11:27:41 140487941920512 [Note] WSREP: Member 1.0 (server.company.local) synced with group.
2019-01-15 11:47:23 140487941920512 [Note] WSREP: Member 1.0 (server.company.local) desyncs itself from group
2019-01-15 11:47:24 140487941920512 [Note] WSREP: Member 1.0 (server.company.local) resyncs itself to group
2019-01-15 11:47:24 140487941920512 [Note] WSREP: Member 1.0 (server.company.local) synced with group.
2019-01-15 12:24:39 140452405958400 [Note] WSREP: MDL BF-BF conflict
schema: databasename
request: (8227134 seqno 46874664 wsrep (2, 1, 0) cmd 3 3 ALTER TABLE `aagenda` ADD `id_subject_cat` int(11) NULL DEFAULT '0' AFTER `id_subject`, ADD INDEX `id_s$
granted: (15 seqno 46874665 wsrep (1, 0, 0) cmd 0 147 (null))
2019-01-15 12:24:40 140452405958400 [Note] WSREP: MDL BF-BF conflict
schema: databasename
request: (8227134 seqno 46874664 wsrep (2, 1, 0) cmd 3 3 ALTER TABLE `aagenda` ADD `id_subject_cat` int(11) NULL DEFAULT '0' AFTER `id_subject`, ADD INDEX `id_s$
granted: (15 seqno 46874665 wsrep (1, 0, 0) cmd 0 147 (null))
2019-01-15 12:24:40 140452405958400 [Note] WSREP: MDL BF-BF conflict
schema: databasename
request: (8227134 seqno 46874664 wsrep (2, 1, 0) cmd 3 3 ALTER TABLE `aagenda` ADD `id_subject_cat` int(11) NULL DEFAULT '0' AFTER `id_subject`, ADD INDEX `id_s$
granted: (11 seqno 46874666 wsrep (1, 0, 0) cmd 0 147 (null))
2019-01-15 12:24:40 0x7fbd9fc3d700 InnoDB: Assertion failure in file /home/buildbot/buildbot/padding_for_CPACK_RPM_BUILD_SOURCE_DIRS_PREFIX/mariadb-10.2.14/storage/innobase/row/row0merge.cc l$
InnoDB: Failing assertion: table->get_ref_count() == 0
InnoDB: We intentionally generate a memory trap.
InnoDB: Submit a detailed bug report to https://jira.mariadb.org/
InnoDB: If you get repeated assertion failures or crashes, even
InnoDB: immediately after the mysqld startup, there may be
InnoDB: corruption in the InnoDB tablespace. Please refer to
InnoDB: https://mariadb.com/kb/en/library/xtradbinnodb-recovery-modes/
InnoDB: about forcing recovery.
190115 12:24:40 [ERROR] mysqld got signal 6 ;
This could be because you hit a bug. It is also possible that this binary
or one of the libraries it was linked against is corrupt, improperly built,or misconfigured. This error can also be caused by malfunctioning hardware.
To report this bug, see https://mariadb.com/kb/en/reporting-bugs
We will try our best to scrape up some info that will hopefully help
diagnose the problem, but since we have already crashed,
something is definitely wrong and this may fail.
Server version: 10.2.14-MariaDB-log
key_buffer_size=134217728
read_buffer_size=131072
max_used_connections=837
max_threads=1502
thread_count=280
It is possible that mysqld could use up to
key_buffer_size + (read_buffer_size + sort_buffer_size)*max_threads = 3431472 K bytes of memory
Hope that's ok; if not, decrease some variables in the equation.
Thread pointer: 0x7fbe2d906c18
Attempting backtrace. You can use the following information to find out
where mysqld died. If you see no messages after this, something went terribly wrong...
stack_bottom = 0x7fbd9fc3cd80 thread_stack 0x49000
/usr/sbin/mysqld(my_print_stacktrace+0x2b)[0x55f4e00d8fab]
/usr/sbin/mysqld(handle_fatal_signal+0x535)[0x55f4dfbad005]
/lib64/libpthread.so.0(+0xf7e0)[0x7fc5f97f67e0]
/lib64/libc.so.6(gsignal+0x35)[0x7fc5f7e50495]
/lib64/libc.so.6(abort+0x175)[0x7fc5f7e51c75]
/usr/sbin/mysqld(+0x47c4eb)[0x55f4df97a4eb]
/usr/sbin/mysqld(+0x90edcc)[0x55f4dfe0cdcc]
/usr/sbin/mysqld(+0x873236)[0x55f4dfd71236]
/usr/sbin/mysqld(_Z17mysql_alter_tableP3THDPcS1_P14HA_CREATE_INFOP10TABLE_LISTP10Alter_infojP8st_orderb+0x29ed)[0x55f4dfab181d]
/usr/sbin/mysqld(_ZN19Sql_cmd_alter_table7executeEP3THD+0x3ae)[0x55f4dfaf62fe]
/usr/sbin/mysqld(_Z21mysql_execute_commandP3THD+0xf81)[0x55f4dfa2b251]
/usr/sbin/mysqld(_Z11mysql_parseP3THDPcjP12Parser_statebb+0x29a)[0x55f4dfa327ca]
/usr/sbin/mysqld(+0x5348c0)[0x55f4dfa328c0]
/usr/sbin/mysqld(_Z16dispatch_command19enum_server_commandP3THDPcjbb+0x18cd)[0x55f4dfa346fd]
/usr/sbin/mysqld(_Z10do_commandP3THD+0x16e)[0x55f4dfa350ee]
/usr/sbin/mysqld(_Z24do_handle_one_connectionP7CONNECT+0x16f)[0x55f4dfaf335f]
/usr/sbin/mysqld(handle_one_connection+0x44)[0x55f4dfaf3484]
/lib64/libpthread.so.0(+0x7aa1)[0x7fc5f97eeaa1]
/lib64/libc.so.6(clone+0x6d)[0x7fc5f7f06bdd]
Trying to get some variables.
Some pointers may be invalid and cause the dump to abort.
Query (0x7fbe2d9141f0): is an invalid pointer
Connection ID (thread ID): 8227134
Status: NOT_KILLED
Optimizer switch: index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_merge_sort_intersection=off,engine_condition_pushdown=off,index_condition_push$
The manual page at http://dev.mysql.com/doc/mysql/en/crashing.html contains
information that should help you find out what is causing the crash.
We think the query pointer is invalid, but we will try to print it anyway.
Query: ALTER TABLE `aagenda` ADD `id_subject_cat` int(11) NULL DEFAULT '0' AFTER `id_subject`, ADD INDEX `id_subject_cat` (`id_subject_cat`)

Answer 1

Advice that I got from MariaDB:
If you want to ALTER a table in a Galera-production environment without downtime do this per node:
SET GLOBAL wsrep_desync = TRUE; (OR SET GLOBAL wsrep_desync = ON;)
SET SESSION wsrep_on = FALSE; (OR SET GLOBAL wsrep_on= OFF ;)
--- ALTER STATEMENT ---
SET SESSION wsrep_on = TRUE; (OR SET GLOBAL wsrep_on= ON ;)
SET GLOBAL wsrep_desync = FALSE; (OR SET GLOBAL wsrep_desync = OFF;)
But the table structure has to be backwards compatible-usable by the application, otherwise you have to stop the cluster and then alter your tables on one node, restart the cluster.