MySQL 5.6.14 Community Release Notes

Thank you to the MySQL Community, on behalf of the MySQL team @ Oracle. Your bug reports, testcases and patches have helped create a better MySQL 5.6.14.

In particular:

  • Thanks to Justin Swanhart for pointing out an issue with statistics on partitioned tables – bug #69179. While we marked Justin’s bug as a duplicate, Justin continued to provide commentary on an earlier bug opened, and we appreciate his feedback.
  • Justin Swanhart also noticed a change in behaviour in MySQL 5.6 with an assertion raised in InnoDB – bug #70007. We’ve since reverted to the previous behaviour.
  • Thanks to Zhai Weixiang, who noticed that InnoDB code intended to only run in debug versions of MySQL was unintentionally running in all versions of mysql. Bug #69724.
  • Thanks to both Robert Lineweaver and Ryan Kenney for reporting bugs #69693 and #69707, as well as providing excellent test cases and research into versions effected. The problems both related to InnoDB mishandling some foreign key names.
  • Thanks to Roel Van de Paar who reported an assertion when innodb_log_group_home_dir does not exist. Bug #69000
  • Thanks to Elena Stepanova for reporting bug #65146. From Elena’s contribution, we both updated the documentation and now produce a warning when START TRANSACTION WITH CONSISTENT SNAPSHOT is used in isolation levels other than REPEATABLE READ.
  • Thanks to Mark Callaghan for reporting an issue with InnoDB’s background thread and shutting down the mysql server. Bug #63276.
  • Thanks to Alexey Kopytov for detecting a situation where an infinite loop could occur in handling InnoDB’s compressed pages. Alexey’s bug report contained detailed code analysis and was very helpful. Bug #61132.
  • Thanks to Shahriyar Rzayev for reporting the original bug, and uploading detailed data to assist in reproduction. It was from these contributions we were later able to devise a testcase and reproduce the bug internally! Bug #69898
  • Thanks to Justin Swanhart and Tsubasa Tanaka who both identified a bug in MASTER_DELAY not type checking input. Bugs #69249 and #69469.
  • Thanks to Hartmut Holzgraefe for reporting an issue with Row-based-replication and decimal data type between MySQL versions. Bug #57166.
  • Thanks to Mark Callaghan for reporting a performance regression in 5.6. Bug #68825
  • Thanks to Jean Weisbuch for identifying an issue with information_schema. Bug #68076.
  • Thanks to Simon Mudd for discovering an issue where the event scheduler did not correctly report when it could not create a new thread. Bug #67191
  • Thanks to Sergey Petrunya who reported bug #67507, as well as the community of users who provided input on the bug – mysqldump is often run from cron, so changes in behaviour risk backups not being performed.
  • Thanks to a private bug reporter who demonstrated that MySQL inadequently protects against stack overflow, and provided an example patch. Bug #35019
  • Thanks to “Dude Letme BE” who reported symbols were missing from libmysql.dll. Bugs #69204, #62394.
  • Thanks to Chito Angeles for reporting a bug in InnoDB fulltext search, and reducing it to a simplified testcase. Bug #69932.

Thank you again all the names listed above. In particular, I would like to call out the two names that appear more than once: Justin Swanhart (3), Mark Callaghan (2).

- Morgan