Saturday, July 30, 2016

Fun with Bugs #44 - Community Bugs Fixed in MySQL 5.7.14

MySQL 5.7.14 was officially released yesterday. So, it's time to check what bugs reported by MySQL Community in public were fixed in this release. Some of these bugs are presented below.

As usual, let me start with InnoDB. The following bugs were fixed there:
  • Bug #80296 - "FTS query exceeds result cache limit". It was reported (for 5.6, but I do not see new 5.6 release notes yet) by Monty Solomon and verified by Umesh.
  • Bug #80304 - "generated columns don't work with foreign key actions". It was reported by Guilhem Bichot based on test case by Peter Gulutzan presented here.As most community bug reports during last 2-3 years, it was verified by Umesh.
  • Bug #80298 - "Full-Text queries with additional secondary index gives NULL or Zero rows", was reported by Ray Lambe and verified by Umesh.
  • Bug #76728 - "reduce lock_sys->mutex contention for transaction that only contains SELECT". This old bug report by Zhai Weixiang (who had provided a patch) was verified by Sinisa Milivojevic.
  • Bug #80083 - "Setting innodb_monitor_enable to ALL does not enable all monitors". It was reported by Davi Arnaut and verified by Miguel Solorzano.
  • Bug #79772 - "Foreign key not allowed when a virtual index exists". It was reported and verified by Jesper wisborg Krogh from Oracle.
There are many more bugs fixed in InnoDB, but all of them were reported in internal Oracle's bugs database by Oracle employees. I do not like this trend.

Now, let's check replication bugs that were fixed:
  • Bug #79324 - "Slave is ~10x slower to execute set of statements compared to master RBR", was reported by Serge Grachov and verified by Umesh.
  • Bug #62008 - "read-only option does not allow inserts/updates on temporary tables". This bug was reported long time ago by Ivan Stanojevic and verified by me when I worked in Oracle.It's really good to see it fixed now!
Some bugs were fixed in Performance_Schema (who could imagine it has bugs...), but they were either reported internally or remain private, like Bug #81464. Just take into account that SELECT from some P_S tables could crash server before 5.7.14, based on release notes...

This time I see several build-related bugs fixed, like these:
  • Bug #81274 - "Add support for Solaris Studio 12.5 aka 5.14". It was reported (and probably fixed) by Tor Didriksen.
  • Bug #81593 - "adapt to gcc 5.3 on solaris". It was also reported and fixed by Tor Didriksen. personally I am happy to see that Oracle still cares about Solaris and related software. Historical sentiments...
  • Bug #80996 - "correct make_pair for c++11 (contribution)". This fix was contributed by Daniel Black
  • Bug #80371 - "MySQL fails to build with new default mode in GCC6". It was reported by Terje Røsten.
The last but not the least, I also have to mention this bug in audit (and, thus, query rewrite) plugins, Bug #81298 - "query rewrite plugin suffers scalability issues". It was reported by Vadim Tkachenko and verified by Sinisa Milivojevic. This is a great improvement.

To summarize, I see reasons to upgrade for those who rely a lot on FTS indexes in InnoDB, replication, audit plugins and Performance_schema. I had not even tried to build 5.7.14 from source yet, so I do not have any personal experience to share.


  1. Adding some links to community bugs where they are missing in the release notes:

    MySQL Server upgrades performed using RPM packages failed when upgrading from MySQL 5.6 Community to MySQL 5.7 Community or MySQL 5.6 Commercial to MySQL 5.7 Commercial. (Bug #23736787)

    Upgrading from native MySQL 5.6 Debian/Ubuntu packages to Oracle-supplied MySQL 5.7 packages caused server restart failures. (Bug #23498230)

    Audit log filtering against the user was performing comparisons against USER(), not CURRENT_USER(). (Bug #23344762)

    After upgrading only the server using RPM packages, the server could fail to start, complaining about a bad errmsg.sys file. (Bug #23338603)

    Upgrading from native Ubuntu 5.7.12 packages to MySQL 5.7.13 packages failed with conflict messages. (Bug #23327563)

    The sys schema format_path() function replaced substrings of the path name argument without verifying that the substrings were delimited by path name separators. Replacement now works correctly, including on Windows. A consequence is that backslashes in Windows path names are no longer converted to forward slashes in the result. (Bug #21512106)

  2. Any reason you didn't include this one? I guess you only focussed on build, replication and InnoDB bugs?

  3. And for those reading the release notes: