Renaud added a comment - 2017-09-07 12:47 We had again the crash this night, please help... Here's the log : 170907 2:20:02 [ERROR] mysqld got exception 0xe06d7363 ; 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: 10.1.22-MariaDB key_buffer_size=134217728 read_buffer_size=131072 max_used_connections=12 max_threads=1001 thread_count=12 It is possible that mysqld could use up to key_buffer_size + (read_buffer_size + sort_buffer_size)*max_threads = 2311984 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... KERNELBASE.dll!RaiseException() mysqld.exe!_CxxThrowException()[throw.cpp:136] mysqld.exe!std::_Xout_of_range()[xthrow.cpp:25] mysqld.exe!dict_stats_analyze_index_for_n_prefix()[dict0stats.cc:1757] mysqld.exe!dict_stats_analyze_index()[dict0stats.cc:2086] mysqld.exe!dict_stats_update_persistent()[dict0stats.cc:2289] mysqld.exe!dict_stats_update()[dict0stats.cc:3239] mysqld.exe!dict_stats_process_entry_from_recalc_pool()[dict0stats_bg.cc:458] mysqld.exe!dict_stats_thread()[dict0stats_bg.cc:551] kernel32.dll!BaseThreadInitThunk() ntdll.dll!RtlUserThreadStart() 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. 2017-09-07 6:07:05 4668 [Note] InnoDB: Using mutexes to ref count buffer pool pages 2017-09-07 6:07:05 4668 [Note] InnoDB: The InnoDB memory heap is disabled 2017-09-07 6:07:05 4668 [Note] InnoDB: Mutexes and rw_locks use Windows interlocked functions 2017-09-07 6:07:05 4668 [Note] InnoDB: _mm_lfence() and _mm_sfence() are used for memory barrier 2017-09-07 6:07:05 4668 [Note] InnoDB: Compressed tables use zlib 1.2.3 2017-09-07 6:07:05 4668 [Note] InnoDB: Using generic crc32 instructions 2017-09-07 6:07:05 4668 [Note] InnoDB: Initializing buffer pool, size = 2.0G 2017-09-07 6:07:05 4668 [Note] InnoDB: Completed initialization of buffer pool 2017-09-07 6:07:06 4668 [Note] InnoDB: Highest supported file format is Barracuda. 2017-09-07 6:07:06 4668 [Note] InnoDB: Starting crash recovery from checkpoint LSN=44449667045 2017-09-07 6:07:06 4668 [Note] InnoDB: Restoring possible half-written data pages from the doublewrite buffer... InnoDB: 4 transaction(s) which must be rolled back or cleaned up InnoDB: in total 4 row operations to undo InnoDB: Trx id counter is 1337856 2017-09-07 6:07:07 4668 [Note] InnoDB: Starting final batch to recover 192 pages from redo log 2017-09-07 6:07:08 4668 [Note] InnoDB: 128 rollback segment(s) are active. InnoDB: Starting in background the rollback of uncommitted transactions 2017-09-07 06:07:08 24c InnoDB: Rolling back trx with id 1337308, 1 rows to undo 2017-09-07 6:07:08 4668 [Note] InnoDB: Waiting for purge to start 2017-09-07 6:07:08 588 [Note] InnoDB: Rollback of trx with id 1337308 completed 2017-09-07 06:07:08 24c InnoDB: Rolling back trx with id 1337306, 1 rows to undo 2017-09-07 6:07:08 588 [Note] InnoDB: Rollback of trx with id 1337306 completed 2017-09-07 06:07:08 24c InnoDB: Rolling back trx with id 1337305, 1 rows to undo 2017-09-07 6:07:08 4668 [Note] InnoDB: Percona XtraDB (http://www.percona.com) 5.6.35-80.0 started; log sequence number 44450925376 2017-09-07 6:07:08 588 [Note] InnoDB: Rollback of trx with id 1337305 completed 2017-09-07 06:07:08 24c InnoDB: Rolling back trx with id 1337214, 1 rows to undo 2017-09-07 6:07:08 588 [Note] InnoDB: Rollback of trx with id 1337214 completed 2017-09-07 06:07:08 24c InnoDB: Rollback of non-prepared transactions completed 2017-09-07 6:07:08 5848 [Note] InnoDB: Dumping buffer pool(s) not yet started 2017-09-07 6:07:08 4668 [Note] Plugin 'FEEDBACK' is disabled. 2017-09-07 6:07:08 4668 [Note] Server socket created on IP: '::'. 2017-09-07 6:07:08 4668 [Note] C:\Program Files\MariaDB 10.1\bin\mysqld.exe: ready for connections. Version: '10.1.22-MariaDB' socket: '' port: 3306 mariadb.org binary distribution

Elena Stepanova added a comment - 2017-09-07 13:51 - edited rcb73 , As our InnoDB expert marko has found out, it is most likely a duplicate of MySQL bug which was fixed here: commit 29acdaaaeef9afe32b42785f1da3d79d56ed7e59 Author: Thirunarayanan Balathandayuthapani <thirunarayanan.balathandayuth@oracle.com> Date: Wed Feb 8 12:00:52 2017 +0530   Bug #24585978 INNODB: ASSERTION TOTAL_RECS > 0 FAILURE IN FILE DICT0STATS.CC This bugfix should be in versions 10.0.31, 10.1.24, 10.2.7, 10.3.1. You have 10.1.22, so please try to upgrade to the latest 10.1 and see if it helps.

Renaud added a comment - 2017-09-07 15:21 Thanks Elena, we have done a lot of tests (regression, performance, etc...) with version 10.1.22, so i was wondering if there's a risk to upgrade from 10.1.22 to 10.1.26 ? Is it only bug fixes ? or is there any evolutions ? For example is there some parameters changes ? (default values change, new parameters, etc...) thanks for your help.

Elena Stepanova added a comment - 2017-09-07 16:31 rcb73 , For defaults, there was one change between these versions, the default value of innodb_empty_free_list_algorithm has changed from BACKOFF to LEGACY . If you are using some dynamic plugins/engines, there could have been changes as well. Naturally, any upgrade involves risks – even if there are no known regressions between versions, there certainly can be hidden ones, maybe something very specific to your particular workflow. However, sooner or later you will have to upgrade – if not on any other reason, then because of security fixes, there have been quite a few already since 10.1.22, and they will keep coming. Given that and the crashes that you encounter, you could just as well start evaluating the new version now.