Details
-
Type:
Bug
-
Status: Closed
-
Priority:
Major
-
Resolution: Duplicate
-
Affects Version/s: 10.0.12-galera
-
Fix Version/s: 10.0.14-galera
-
Component/s: Galera
-
Labels:
-
Environment:- RHEL 6.5
- MGC 10.0.12
Description
# my_print_defaults mysqld --performance_schema=on --datadir=/opt/data/mysql --tmpdir=/tmp/ --log_error=/opt/data/mysql/error.log --log_warnings=2 --query_cache_size=0 --query_cache_type=0 --log_bin --log_slave_updates --binlog_format=ROW --innodb_log_file_size=128M --innodb_buffer_pool_size=2G --innodb_flush_method=O_DIRECT --innodb_flush_log_at_trx_commit=1 --innodb_autoinc_lock_mode=2 --wsrep_cluster_name=GaleraPOC --wsrep_cluster_address=gcomm://mariadb001,mariadb002,mariadb003 --wsrep_provider=/usr/lib64/galera/libgalera_smm.so --wsrep_provider_options=gcache.size=512M --wsrep_sst_method=xtrabackup-v2 --wsrep_sst_auth=sst:sstpass --server_id=1 --wsrep_node_name=mariadb001 --wsrep_node_address=mariadb001 --wsrep_sst_receive_address=mariadb001
* 10.0.12-MariaDB-wsrep-log MariaDB Server, wsrep_25.10.r4002 * wsrep_provider_version=25.3.5(rXXXX)
# ./node1.sh .............ERROR 2013 (HY000) at line 1: Lost connection to MySQL server during query ERROR with query
# ./node2.sh .......
The server on node1 crashes:
140828 15:34:49 [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 http://kb.askmonty.org/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: 10.0.12-MariaDB-wsrep-log key_buffer_size=134217728 read_buffer_size=131072 max_used_connections=7 max_threads=153 thread_count=8 It is possible that mysqld could use up to key_buffer_size + (read_buffer_size + sort_buffer_size)*max_threads = 467215 K bytes of memory Hope that's ok; if not, decrease some variables in the equation. Thread pointer: 0x0x7f2d7ce8f008 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 = 0x7f2e43e56ce0 thread_stack 0x48000 /usr/sbin/mysqld(my_print_stacktrace+0x2b)[0xb914fb] /usr/sbin/mysqld(handle_fatal_signal+0x398)[0x743318] /lib64/libpthread.so.0(+0xf710)[0x7f2e43b1c710] /lib64/libc.so.6(+0x83742)[0x7f2e421cc742] /usr/sbin/mysqld(_ZNK19rpl_sql_thread_info22cached_charset_compareEPc+0x20)[0x699bf0] /usr/sbin/mysqld(_ZN15Query_log_event14do_apply_eventEP14rpl_group_infoPKcj+0x837)[0x800817] /usr/sbin/mysqld(_Z14wsrep_apply_cbPvPKvmjPK14wsrep_trx_meta+0x525)[0x6f05e5] /usr/lib64/galera/libgalera_smm.so(_ZNK6galera9TrxHandle5applyEPvPF15wsrep_cb_statusS1_PKvmjPK14wsrep_trx_metaERS6_+0xb1)[0x7f2e3eb 4c2c1] /usr/lib64/galera/libgalera_smm.so(+0x1aaf95)[0x7f2e3eb83f95] /usr/lib64/galera/libgalera_smm.so(_ZN6galera13ReplicatorSMM10replay_trxEPNS_9TrxHandleEPv+0x12e)[0x7f2e3eb8485e] /usr/lib64/galera/libgalera_smm.so(galera_replay_trx+0x5c)[0x7f2e3eb9845c] /usr/sbin/mysqld(_Z24wsrep_replay_transactionP3THD+0x2de)[0x6f217e] /usr/sbin/mysqld[0x5e1120] /usr/sbin/mysqld(_Z16dispatch_command19enum_server_commandP3THDPcj+0x16d0)[0x5e28a0] /usr/sbin/mysqld(_Z10do_commandP3THD+0x132)[0x5e3072] /usr/sbin/mysqld(_Z24do_handle_one_connectionP3THD+0x54b)[0x6a190b] /usr/sbin/mysqld(handle_one_connection+0x42)[0x6a1a02] /lib64/libpthread.so.0(+0x79d1)[0x7f2e43b149d1] /lib64/libc.so.6(clone+0x6d)[0x7f2e42231b5d] Trying to get some variables. Some pointers may be invalid and cause the dump to abort. Query (0x7f2e1b631251): is an invalid pointer Connection ID (thread ID): 331 Status: NOT_KILLED
I spoke with Teemu from Codership and he mentioned this is likely a bug specific to MariaDB.
I tested this on:
- 5.5.38: no crash
- PXC 5.6.19: no crash
12:22 < gryp> I have crash of MGC-10.0.12: https://gist.github.com/grypyrg/22b0512cb59b5e32a538, seems to be similar to
knielsen's fix of https://mariadb.atlassian.net/browse/MDEV-6156 . Anybody got any ideas?
12:27 < knielsen> gryp: The code path involved is completely different (Galera replication vs. MariaDB parallel replication).
But it is possible that Galera has a similar bug (it also does parallel replication), might help the
developers track it down...
12:29 < knielsen> Interesting that Galera seems to be calling Query_log_event::do_apply_event() ... I suppose this is to
handle DDL perhaps, which shouldn't run in parallel
12:29 < knielsen> well, not that I know much of what Galera code does
12:29 < gryp> indeed. it's during some DDL's. I'm still working on getting a cleaner test case.
12:31 < gryp> I'm also running with wsrep_slave_threads=1 and the DDL is executed on the crashing node.
12:31 < gryp> knielsen: I'll talk some more with Codership, see what they think based on your feedback. TNx
12:33 < knielsen> gryp: there's more discussions of this bug on maria-developers@ (google should know), in case it helps
understand the issue. Though it could be something different, hard to say
12:33 < gryp> Yep, have found that as well. tnx. (https://www.mail-archive.com/maria-developers@lists.launchpad.net/msg06846.html)
Gliffy Diagrams
Attachments
Activity
- All
- Comments
- Work Log
- History
- Activity
- Transitions