[MDEV-25118] Galera : Crash after node recovery Created: 2021-03-11  Updated: 2021-12-21  Resolved: 2021-04-13

Status: Closed
Project: MariaDB Server
Component/s: Galera
Affects Version/s: 10.4.18
Fix Version/s: 10.4.19

Type: Bug Priority: Major
Reporter: Yoann Laissus Assignee: Jan Lindström (Inactive)
Resolution: Fixed Votes: 1
Labels: None
Environment:

Debian 10 with official repositories on a KVM VM

Attachments: HTML File config     File galera01.log     File galera02.log     File galera03.log    
Issue Links:
Relates
relates to MDEV-24878 Signal 11 on TABLE_LIST::placeholder() Closed

 Description   

We are using a three node Galera cluster with MariaDB 10.4.18 from official repositories.
The three nodes are using the same configuration.

Since some weeks, we are facing some crashes after a node revival.
It's not reproducable every time unfortunately.
I don't know the exact version introducing this problem, it happened probably after our migration to 10.4.x)

I uploaded a crash we had yesterday. The backtrace seems to indicate it's related to table locking.

How it happened :

  • In the morning we stopped galera01 for some maintenance operations on the underlying host.
  • The cluster was working fine the whole day with only two nodes (galera02 et galera03).
  • When we started again galera01 at 22:50:02, the donor node (galera02) crashed after some seconds.
  • Then, all the three nodes started to crash in a loop until stabilize them after some minutes.

Some other times, the whole cluster totally crashed and we need to restart it manually.
The backtrace of the crash is similar to the other cases.

I can add some more debug messages if needed.

The full backtrace :

210310 22:50:03 [ERROR] mysqld got signal 11 ;
 
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: 10.4.18-MariaDB-1:10.4.18+maria~buster
 
key_buffer_size=16777216
 
read_buffer_size=131072
 
max_used_connections=5
 
max_threads=153
 
thread_count=15
 
It is possible that mysqld could use up to 
key_buffer_size + (read_buffer_size + sort_buffer_size)*max_threads = 353072 K  bytes of memory
 
Hope that's ok; if not, decrease some variables in the equation.
 
 
 
Thread pointer: 0x7f9660001268
 
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 = 0x7f9824123e08 thread_stack 0x40000
 
/usr/sbin/mysqld(my_print_stacktrace+0x2e)[0x55d33bb277fe]
 
/usr/sbin/mysqld(handle_fatal_signal+0x54d)[0x55d33b61ab0d]
 
/lib/x86_64-linux-gnu/libpthread.so.0(+0x12730)[0x7f9850da8730]
 
/usr/sbin/mysqld(+0x64819d)[0x55d33b3aa19d]
 
/usr/sbin/mysqld(_Z11lock_tablesP3THDP10TABLE_LISTjj+0x4a8)[0x55d33b3aff78]
 
/usr/sbin/mysqld(_Z20open_and_lock_tablesP3THDRK14DDL_options_stP10TABLE_LISTbjP19Prelocking_strategy+0x8a)[0x55d33b3b16ba]
 
/usr/sbin/mysqld(_Z24open_n_lock_single_tableP3THDP10TABLE_LIST13thr_lock_typejP19Prelocking_strategy+0x41)[0x55d33b3b1761]
 
/usr/sbin/mysqld(+0x8370c5)[0x55d33b5990c5]
 
/usr/sbin/mysqld(_ZNK12Wsrep_schema12restore_viewEP3THDRKN5wsrep2idE+0xf8)[0x55d33b59bd88]
 
/usr/sbin/mysqld(_ZN20Wsrep_server_service8log_viewEPN5wsrep21high_priority_serviceERKNS0_4viewE+0x4d4)[0x55d33b7970d4]
 
/usr/sbin/mysqld(_ZN5wsrep12server_state7on_viewERKNS_4viewEPNS_21high_priority_serviceE+0x39b)[0x55d33bb9f95b]
 
/usr/sbin/mysqld(+0xe4facc)[0x55d33bbb1acc]
 
/usr/lib/galera/libgalera_smm.so(+0x5ccd9)[0x7f984d5accd9]
 
/usr/lib/galera/libgalera_smm.so(+0x5cd18)[0x7f984d5acd18]
 
/usr/lib/galera/libgalera_smm.so(+0x68398)[0x7f984d5b8398]
 
/usr/lib/galera/libgalera_smm.so(+0x685ae)[0x7f984d5b85ae]
 
/usr/lib/galera/libgalera_smm.so(+0x92dd8)[0x7f984d5e2dd8]
 
/usr/lib/galera/libgalera_smm.so(+0x93061)[0x7f984d5e3061]
 
/usr/lib/galera/libgalera_smm.so(+0x66cb0)[0x7f984d5b6cb0]
 
/usr/lib/galera/libgalera_smm.so(+0x4248b)[0x7f984d59248b]
 
/usr/sbin/mysqld(_ZN5wsrep18wsrep_provider_v2611run_applierEPNS_21high_priority_serviceE+0xe)[0x55d33bbaf4fe]
 
/usr/sbin/mysqld(+0x836dff)[0x55d33b598dff]
 
/usr/sbin/mysqld(_Z15start_wsrep_THDPv+0x388)[0x55d33b58aac8]
 
/lib/x86_64-linux-gnu/libpthread.so.0(+0x7fa3)[0x7f9850d9dfa3]
 
/lib/x86_64-linux-gnu/libc.so.6(clone+0x3f)[0x7f98502b24cf]
 
 
 
Trying to get some variables.
 
Some pointers may be invalid and cause the dump to abort.
 
Query (0x0): (null)
 
Connection ID (thread ID): 17
 
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_pushdown=on,derived_merge=on,derived_with_keys=on,firstmatch=on,loosescan=on,materialization=on,in_to_exists=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on,subquery_cache=on,mrr=off,mrr_cost_based=off,mrr_sort_keys=off,outer_join_with_cache=on,semijoin_with_cache=on,join_cache_incremental=on,join_cache_hashed=on,join_cache_bka=on,optimize_join_buffer_size=on,table_elimination=on,extended_keys=on,exists_to_in=on,orderby_uses_equalities=on,condition_pushdown_for_derived=on,split_materialized=on,condition_pushdown_for_subquery=on,rowid_filter=on,condition_pushdown_from_having=on
 
 
 
The manual page at https://mariadb.com/kb/en/how-to-produce-a-full-stack-trace-for-mysqld/ 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: 
 
 
Writing a core file...
 
Working directory at /var/lib/mysql
 
Resource Limits:
 
Limit                     Soft Limit           Hard Limit           Units     
Max cpu time              unlimited            unlimited            seconds   
Max file size             unlimited            unlimited            bytes     
Max data size             unlimited            unlimited            bytes     
Max stack size            8388608              unlimited            bytes     
Max core file size        0                    unlimited            bytes     
Max resident set          unlimited            unlimited            bytes     
Max processes             38887                38887                processes 
Max open files            1048576              1048576              files     
Max locked memory         65536                65536                bytes     
Max address space         unlimited            unlimited            bytes     
Max file locks            unlimited            unlimited            locks     
Max pending signals       38887                38887                signals   
Max msgqueue size         819200               819200               bytes     
Max nice priority         0                    0                    
Max realtime priority     0                    0                    
Max realtime timeout      unlimited            unlimited            us        
Core pattern: core



 Comments   
Comment by Frank Groot [ 2021-03-25 ]

We are seeing the same issue. Only difference is timing.
At the moment we bring down node-1, node-3 logs forgetting node-1 and then crashes with signal 11 and the same stacktrace.

Server version: 10.4.18-MariaDB-1:10.4.18+maria~focal-log

Comment by Andreas Hering [ 2021-04-12 ]

We got that problems too. Since 10.4.18 we see crashes without any particular reason. This happens as @FrankGroot described when one node was brought down. And just after disabling one node another node just crashes with signal 11.
We couldnt reproduce the behaviour, as the cluster consists only of 3 nodes and is a live cluster. Our test environment does not have this issues. Maybe it is linked to the usage as it seems to be refering to locking of tables (see log part).

Version: mariadb Ver 15.1 Distrib 10.4.18-MariaDB, for Linux (x86_64) using readline 5.1
OS: CentOS Linux release 7.9.2009

210412 10:49:57 [ERROR] mysqld got signal 11 ;
 
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: 10.4.18-MariaDB-log
 
key_buffer_size=134217728
 
read_buffer_size=131072
 
max_used_connections=10
 
max_threads=1026
 
thread_count=19
 
It is possible that mysqld could use up to
 
key_buffer_size + (read_buffer_size + sort_buffer_size)*max_threads = 2389008 K  bytes of memory
 
Hope that's ok; if not, decrease some variables in the equation.
 
 
 
Thread pointer: 0x7f99800009a8
 
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 = 0x7f9b20090d20 thread_stack 0x49000
 
/usr/sbin/mysqld(my_print_stacktrace+0x2e)[0x55aaa379a47e]
 
/usr/sbin/mysqld(handle_fatal_signal+0x30f)[0x55aaa3228e8f]
 
sigaction.c:0(__restore_rt)[0x7f9b26521630]
 
/usr/sbin/mysqld(+0x655d49)[0x55aaa2fb0d49]
 
/usr/sbin/mysqld(_Z11lock_tablesP3THDP10TABLE_LISTjj+0x46d)[0x55aaa2fb6e3d]
 
/usr/sbin/mysqld(_Z20open_and_lock_tablesP3THDRK14DDL_options_stP10TABLE_LISTbjP19Prelocking_strategy+0x8a)[0x55aaa2fb846a]
 
/usr/sbin/mysqld(_Z24open_n_lock_single_tableP3THDP10TABLE_LIST13thr_lock_typejP19Prelocking_strategy+0x41)[0x55aaa2fb8511]
 
/usr/sbin/mysqld(+0x84201b)[0x55aaa319d01b]
 
/usr/sbin/mysqld(+0x8422a1)[0x55aaa319d2a1]
 
/usr/sbin/mysqld(_ZNK12Wsrep_schema12restore_viewEP3THDRKN5wsrep2idE+0x553)[0x55aaa319f943]
 
/usr/sbin/mysqld(_ZN20Wsrep_server_service8log_viewEPN5wsrep21high_priority_serviceERKNS0_4viewE+0x57a)[0x55aaa33b2aaa]
 
/usr/sbin/mysqld(_ZN5wsrep12server_state7on_viewERKNS_4viewEPNS_21high_priority_serviceE+0x3f5)[0x55aaa382c1e5]
 
/usr/sbin/mysqld(+0xee10b0)[0x55aaa383c0b0]
 
src/replicator_smm.cpp:2429(galera::ReplicatorSMM::submit_view_info(void*, wsrep_view_info const*))[0x7f9b21076359]
 
src/replicator_smm.cpp:2824(galera::ReplicatorSMM::submit_ordered_view_info(void*, wsrep_view_info const*))[0x7f9b210773f3]
 
src/replicator_smm.cpp:3010(galera::ReplicatorSMM::process_prim_conf_change(void*, gcs_act_cchange const&, int, void*))[0x7f9b21083610]
 
src/galera_gcs.hpp:173(galera::Gcs::resume_recv())[0x7f9b2108381f]
 
src/gcs_action_source.cpp:155(galera::GcsActionSource::dispatch(void*, gcs_action const&, bool&))[0x7f9b210acedd]
 
src/gcs_action_source.cpp:29(~Release)[0x7f9b210ad384]
 
src/replicator_smm.cpp:390(galera::ReplicatorSMM::async_recv(void*))[0x7f9b21081e7b]
 
src/wsrep_provider.cpp:263(galera_recv)[0x7f9b210602b8]
 
/usr/sbin/mysqld(_ZN5wsrep18wsrep_provider_v2611run_applierEPNS_21high_priority_serviceE+0xe)[0x55aaa38399de]
 
/usr/sbin/mysqld(+0x841b52)[0x55aaa319cb52]
 
/usr/sbin/mysqld(_Z15start_wsrep_THDPv+0x377)[0x55aaa318f247]
 
pthread_create.c:0(start_thread)[0x7f9b26519ea5]
 
/lib64/libc.so.6(clone+0x6d)[0x7f9b248b99fd]

Comment by Yoann Laissus [ 2021-04-12 ]

Since this report, we upgraded to 10.5 and it fixed all our issues about that.

It's fixed since 10.5.9 (but not in the 10.4 branch) :
MDEV-24878
https://github.com/MariaDB/server/commit/ae7989ca2059869f81c837509e5ff554f7f63562

Comment by Jan Lindström (Inactive) [ 2021-04-13 ]

yoannl Fix has been already ported to 10.4 also.

Comment by Agbo Steven [ 2021-04-30 ]

Hi,

We are also impacted by this issue.

Do you have ETA for version 10.4.19?

We have planned to upgrade to 10.5.X on Q4 this year and before would like to have this fix (10.4.19)

Comment by Aurélien LEQUOY [ 2021-12-21 ]

got the same error in 10.6.4 after rm /var/lib/mysql need to delete /tmp/galera.cache too

Generated at Thu Feb 08 09:35:17 UTC 2024 using Jira 8.20.16#820016-sha1:9d11dbea5f4be3d4cc21f03a88dd11d8c8687422.