Bug #115321 | query hangs in "query end" state too long after killing | ||
---|---|---|---|
Submitted: | 13 Jun 2024 14:05 | Modified: | 13 Jun 2024 15:32 |
Reporter: | Eugeny Nt | Email Updates: | |
Status: | Not a Bug | Impact on me: | |
Category: | MySQL Server | Severity: | S3 (Non-critical) |
Version: | 8.2.0 | OS: | Windows (11) |
Assigned to: | CPU Architecture: | Any | |
Tags: | innodb, KILL, temporary table |
[13 Jun 2024 14:05]
Eugeny Nt
[13 Jun 2024 14:54]
MySQL Verification Team
Hi Mr. Nt, Thank you for your bug report. However, this is not a bug. There are many places in InnoDB code, where the KILL flag is not checked at all, because killing query would result in destroyed integrity of the data. This is common in all transactional engines. MyISAM is NOT a transactional engine. This is documented in several places, like: https://dev.mysql.com/doc/refman/8.4/en/kill.html (INSERT is not mentioned in the above page) Also here: https://dev.mysql.com/doc/refman/8.4/en/sys-innodb-lock-waits.html Then, there is a warning that KILL should not be used on InnoDB tables: https://dev.mysql.com/doc/refman/8.4/en/crashing.html Then, you have some explanations here: https://dev.mysql.com/doc/refman/8.4/en/optimizing-innodb-transaction-management.html This is simply a known and documented behaviour. Not a bug.
[13 Jun 2024 15:32]
Eugeny Nt
Hi, Team. Thank you for your answer, but I can't find anything related to this problem from your links. These views are empty for me: SELECT * from sys.innodb_lock_waits; SELECT * from sys.x$innodb_lock_waits; And I am inserting to a temporary table, it's not visible for other sessions, so I do not understand how it can be related to transactions?
[14 Jun 2024 9:25]
MySQL Verification Team
Hi Mr. Nt, Those views can be empty for you. However, we did not recommend you to run those views. We just recommended you to read those pages , to show that this is all documented behaviour. Also, in 8.0 and above, temporary tables are based on the InnoDB storage engine. Not a bug.