Details
-
Bug
-
Status: Closed (View Workflow)
-
Critical
-
Resolution: Duplicate
-
10.3.9
-
None
-
CentOS 7 on VMWare esx
32GB RAM
8GB Swap
8vCPU
332GB disk usage
About 150 databases with each about 160 tables - all InnoDB
Description
After upgrading from MariaDB 10.1.32 to MariaDB 10.3.9 on one production server (several others did not encounter this problem), the server failed to start with the following problem:
2018-09-26 01:40:31 0x7f1a0dc088c0 InnoDB: Assertion failure in file /home/buildbot/buildbot/padding_for_CPACK_RPM_BUILD_SOURCE_DIRS_PREFIX/mariadb-10.3.9/storage/innobase/pars/pars0pars.cc line 818
InnoDB: Failing assertion: sym_node->table != NULL
We managed to start MariaDB by using "innodb_force_recovery = 3" and move all data to other database servers. However a drop database command causes the same issue to appear, even with the recovery on 3.
A "mysqlcheck --all-databases" did not find any issues and a mysqldump of all databases also did not cause any problems. This is the reason we think this might be a bug and not just data corruption.
Enclosed are 2 problem traces. The MariaDB-pars0pars.cc-818-first.txt contains the problem just after starting the server in 3.1.9. The second contains the problem with innodb_force_recovery = 3 after a database drop was executed. It failed a few times after the database drop and then started again correctly.
Attachments
Issue Links
- relates to
-
-
- Closed
-
-
MDEV-14585 Automatically remove #sql- tables in innodb dictionary during recovery
-
- Closed
-
Activity
The same failure happened in buildbot (once so far):
http://buildbot.askmonty.org/buildbot/builders/kvm-rpm-centos74-amd64-debug/builds/912
|
10.3 7aba6f8f8853acd18d471793f8b72aa1412b8151 |
innodb_fts.sync_ddl 'innodb' w2 [ fail ]
|
Test ended at 2018-09-22 21:25:16
|
|
|
CURRENT_TEST: innodb_fts.sync_ddl
|
mysqltest: At line 57: query 'TRUNCATE TABLE t1' failed: 2013: Lost connection to MySQL server during query
|
|
|
The result from queries just before the failure was:
|
< snip >
|
DROP TABLE t1;
|
SET GLOBAL debug_dbug = @save_debug;
|
CREATE TABLE t1 (
|
id INT AUTO_INCREMENT NOT NULL PRIMARY KEY,
|
value VARCHAR(1024)
|
) ENGINE=InnoDB;
|
CREATE FULLTEXT INDEX idx1 ON t1(value);
|
SET GLOBAL debug_dbug = '+d,fts_instrument_sync_request,fts_instrument_write_words_before_select_index,ib_trunc_sleep_before_fts_cache_clear';
|
INSERT INTO t1 (value) VALUES
|
('By default or with the IN NATURAL LANGUAGE MODE modifier'),
|
('performs a natural language search for a string'),
|
('collection is a set of one or more columns included'),
|
('returns a relevance value; that is, a similarity measure'),
|
('and the text in that row in the columns named in'),
|
('By default, the search is performed in case-insensitive'),
|
('sensitive full-text search, use a binary collation '),
|
('example, a column that uses the latin1 character'),
|
('collation of latin1_bin to make it case sensitive')
|
;
|
TRUNCATE TABLE t1;
|
|
|
More results from queries before failure can be found in /dev/shm/var/2/log/sync_ddl.log
|
|
|
|
|
Server [mysqld.1 - pid: 30449, winpid: 30449, exit: 256] failed during test run
|
Server log from this test:
|
----------SERVER LOG START-----------
|
2018-09-22 21:25:14 0x7f772affd700 InnoDB: Assertion failure in file /home/buildbot/buildbot/padding_for_CPACK_RPM_BUILD_SOURCE_DIRS_PREFIX/mariadb-10.3.10/storage/innobase/pars/pars0pars.cc line 818
|
InnoDB: Failing assertion: sym_node->table != NULL
|
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.
|
180922 21:25:14 [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: 10.3.10-MariaDB-debug-log
|
key_buffer_size=1048576
|
read_buffer_size=131072
|
max_used_connections=3
|
max_threads=153
|
thread_count=9
|
It is possible that mysqld could use up to
|
key_buffer_size + (read_buffer_size + sort_buffer_size)*max_threads = 63258 K bytes of memory
|
Hope that's ok; if not, decrease some variables in the equation.
|
|
|
Thread pointer: 0x0
|
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 = 0x0 thread_stack 0x49000
|
/usr/sbin/mysqld(my_print_stacktrace+0x2b)[0x562bab906cbf]
|
/usr/sbin/mysqld(handle_fatal_signal+0x33e)[0x562bab238e1e]
|
sigaction.c:0(__restore_rt)[0x7f7748bd35e0]
|
:0(__GI_raise)[0x7f77470e01f7]
|
:0(__GI_abort)[0x7f77470e18e8]
|
/usr/sbin/mysqld(+0xbf324d)[0x562bab64e24d]
|
/usr/sbin/mysqld(+0xac6433)[0x562bab521433]
|
ut/ut0dbg.cc:62(_GLOBAL__sub_I_ut0dbg.cc)[0x562bab525c0f]
|
pars/pars0pars.cc:1321(pars_insert_statement(sym_node_t*, void*, sel_node_t*))[0x562bab7c4f9e]
|
innobase/pars0grm.y:375(yyparse())[0x562bab5292c5]
|
fts/fts0sql.cc:184(fts_parse_sql(fts_table_t*, pars_info_t*, char const*))[0x562bab7a5a2e]
|
fts/fts0fts.cc:3935(fts_write_node(trx_t*, que_fork_t**, fts_table_t*, fts_string_t*, fts_node_t*))[0x562bab7844c4]
|
fts/fts0fts.cc:4070(fts_sync_index(fts_sync_t*, fts_index_cache_t*))[0x562bab78496a]
|
fts/fts0fts.cc:4396(fts_sync(fts_sync_t*, bool, bool, bool))[0x562bab78e6bb]
|
fts/fts0fts.cc:4481(fts_sync_table(dict_table_t*, bool, bool, bool))[0x562bab791235]
|
fts/fts0opt.cc:2831(fts_optimize_sync_table(unsigned long))[0x562bab79382d]
|
fts/fts0opt.cc:2945(fts_optimize_thread(void*))[0x562bab79a8dc]
|
pthread_create.c:0(start_thread)[0x7f7748bcbe25]
|
/lib64/libc.so.6(clone+0x6d)[0x7f77471a334d]
|
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.
|
Writing a core file at /dev/shm/var/2/mysqld.1/data/
|
----------SERVER LOG END-------------
|
|
|
|
|
- found 'core.30450' (0/0)
|
|
|
Trying 'dbx' to get a backtrace
|
gdb not found, cannot get the stack trace
|
|
|
Trying 'lldb' to get a backtrace from coredump /dev/shm/var/2/log/innodb_fts.sync_ddl-innodb/mysqld.1/data/core.30450
|
Compressed file /dev/shm/var/2/log/innodb_fts.sync_ddl-innodb/mysqld.1/data/core.30450
|
- skipping '/dev/shm/var/2/log/innodb_fts.sync_ddl-innodb/'
|
I believe that this problem was emphasized in MDEV-14585 and fixed in MDEV-12023.
The failure in the description could also be possible in MariaDB 10.2 before the MDEV-12023 fix, because the SYS_VIRTUAL table would not exist when dropping orphan internal tables for fulltext indexes.
With MDEV-14585, also tables whose name starts with #sql will be dropped on startup. And this would cause the reported assertion failure if any of the non-hard-coded metadata tables does not exist. On upgrade, the tables would be created a little later.
In MariaDB-pars0pars.cc-818-first.txt
, InnoDB crashes on startup because a data dictionary table is not found while trying to drop a table whose name starts with #sql. This is clearly due to code that was added in MDEV-14585.
In MariaDB-pars0pars.cc-818-second.txt, InnoDB refused to create the data dictionary tables, because innodb_force_recovery was set to 3. So, any DDL statement that would drop tables (DROP TABLE, DROP DATABASE, some ALTER TABLE or OPTIMIZE TABLE) would crash due to the missing data dictionary tables.
I believe that both cases should be fixed by MDEV-12023.
stijn, if you still have a copy of the dataset, could you try an upgrade to a recent 10.3 development snapshot and confirm that the problem has been fixed?
The upgrade to the development snapshot seems to have fixed the issue, thanks!
What kind of upgrade did you perform? Was it a "live" upgrade (shutdown and uninstall the old server, install the new server, and start it on the same datadir), or did you use a backup tool, e.g. XtraBackup or MariaBackup?