Note that recently most of the bugs fixed are internal only, found by Oracle engineers that never cared (or are not allowed, who knows) to report them in public, so this blog post is not even remotely a full review of what's fixed in MySQL 5.7.18 and is not a replacement for reading the detailed release notes.
So, here is the result. I've noted the following InnoDB bugs from Community users are fixed (as usual, I explicitly state who reported and verified the bug):
- Bug #83245 - "Log parsing buffer overflow". This regression bug was reported by my former colleague from Percona, Tomislav Plavcic (based on the original findings by Peter Zaitsev), and verified by Umesh Shastry.
- Bug #82968 - "InnoDB's estimate for rows-in-table can return rows=1 due to a race condition", was reported by my colleague Sergey Petrunya while working on my MariaDB bug report, MDEV-10649 (based, in turn, on a problem one of our key customers hit in production). The bug was kindly verified by Sinisa Milivojevic.
- Bug #80060 - "Concurrent TRUNCATE TABLEs cause stalls". It's great to see this case reported by Domas Mituzas fixed at least in 5.7.x. But it's even better to just read all the comments and arguments in that great and funny discussion he had with Sinisa Milivojevic who was working on verification...
- Bug #80580 - "count(*) much slower on 5.7 than 5.6". This regression (not tagged as such!) bug was reported by Qi Xiaobin as optimizer one (but it turned out to be in InnoDB), and verified by Umesh Shastry. Simple public test case there was provided by me.
- Bug #84202 is not public, so let me simply quote (after telling you again that I hate such cases):
row_search_mvcc()function unnecessarily traversed the entire table for a range query, which occurred when the record was not in the transaction read view. (Bug #84202, Bug #23481444, Bug #25251375)"
- Bug #83537 - "WAIT_UNTIL_SQL_THREAD_AFTER_GTIDS timeout value handles some inputs badly". It was reported by Dan Reif and verified by Umesh Shastry. Now fractional and negative timeouts are processed in a reasonable and expected way.
- Bug #84674 - "Having an unresolvable hostname in group_repl should not block group replication", by Kenny Gryp. I've already reviewed active group replication bugs in a separate post some time ago. Nice to see some of them fixed. The bug was probably verified by Nuno Carvalho.
- Bug #83918 - "RBR: Wrong field length in error message". It was reported by Luis Soares from Oracle, who probably fixed it himself. Nice example when Oracle engineers do not hesitate to report bugs in public!
- Bug #83270 - "Performance degradation when the server has many gtids", was reported by Daniël van Eeden and quickly verified by Umesh Shastry.
- Bug #84786 - "--performance-schema-digests-size=1 leads to SIGSEGV". It was reported by Roel Van de Paar and promptly verified by Miguel Solorzano.
- Bug #84263 - "mysql.server (MySQL Server Startup Script) can not work,and export some error.". This bug was reported by Song Boomballa. See also this my related bug report, Bug #84173 - "mysqld_safe --no-defaults & silently does NOT work any more". Nice to know this regression is finally fixed.
- Bug #84172 - "The pid-file value is ignored in the /etc/my.cnf option file". This nice packaging bug/difference was noted and reported by Monty Solomon.
- Bug #83253 - "performance_schema digests the same query to multiple different digests", was reported by my former colleague Justin Swanhart. The bug was verified and probably fixed by Mayank Prasad.
- Bug #83110 - "found_rows() returns 1 when no rows found". This nice regression bug was found and reported by Yamada Isami, and verified by Umesh Shastry.
- Bug #83005 - "When the Optimiser is using index for group-by it often gives wrong results", was reported by Yoseph Phillips and verified, again by Umesh Shastry. Somehow this regression affecting versions 5.6+ was not noted for a long time.
- Bug #82313 - "Wrong result for EXPLAIN SELECT COUNT(*) FROM TABLE;". This is yet another optimizer regression bug in 5.7 found by Justin Swanhart. I wonder if MySQL 8 (that seemed to be affected) would include the fix... Release notes fro 8.0.1 are empty at the moment!
- Bug #81854 - "Force index is skipped while executing select count(*)", was reported by Zhai Weixiang and verified by Umesh Shastry.
- Bug #78244 - "SELECT DISTINCT, wrong results combined with use_index_extensions=off". It was reported by Daniel G and verified by Miguel Solorzano.
- Bug #41908 - "Trying to start MySQL while another instance is starting: confusing error messag". The bug was reported by Roel Van de Paar more than 8 years ago! It was verified by Sveta Smirnova. Somehow it seems that 5.7.18 had introduced a lot of changes into the way mysqld_safe works...
openxs@ao756:~/dbs/5.7$ bin/mysqld_safe --no-defaults --port=3308 --socket=/tmp/mysql57.sock --basedir=/home/openxs/dbs/5.7 --datadir=/home/openxs/dbs/5.7/data &Time to start testing and looking for something new...
openxs@ao756:~/dbs/5.7$ 2017-04-11T10:51:24.224518Z mysqld_safe Logging to '/home/openxs/dbs/5.7/data/ao756.err'.
2017-04-11T10:51:24.264873Z mysqld_safe Starting mysqld daemon with databases from /home/openxs/dbs/5.7/data
openxs@ao756:~/dbs/5.7$ bin/mysql -uroot --socket=/tmp/mysql57.sock
Welcome to the MySQL monitor. Commands end with ; or \g.
Your MySQL connection id is 3
Server version: 5.7.18 MySQL Community Server (GPL)
Copyright (c) 2000, 2017, Oracle and/or its affiliates. All rights reserved.
Oracle is a registered trademark of Oracle Corporation and/or its
affiliates. Other names may be trademarks of their respective
Type 'help;' or '\h' for help. Type '\c' to clear the current input statement.