Details
-
Technical task
-
Status: Open (View Workflow)
-
Major
-
Resolution: Unresolved
-
10.2(EOL), 10.3(EOL), 10.4(EOL)
-
None
Description
10.2 c17b0b73 |
2019-07-08 1:39:49 139741663909696 [Note] mysqld: Aria engine: starting recovery
|
recovered pages: 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% (1.2 seconds); transactions to roll back: 1
|
Got error 22 when executing undo undo_row_insert
|
2019-07-08 1:39:51 139741663909696 [ERROR] mysqld: Aria engine: Undo phase failed
|
tables to flush: 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0
|
2019-07-08 1:39:51 139741663909696 [ERROR] mysqld: Aria recovery failed. Please run aria_chk -r on all Aria tables and delete all aria_log.######## files
|
2019-07-08 1:39:51 139741663909696 [ERROR] Plugin 'Aria' init function returned error.
|
2019-07-08 1:39:51 139741663909696 [ERROR] Plugin 'Aria' registration as a STORAGE ENGINE failed.
|
2019-07-08 1:39:51 139741663909696 [Note] Plugin 'FEEDBACK' is disabled.
|
2019-07-08 1:39:51 139741663909696 [ERROR] Aria engine is not enabled or did not start. The Aria engine must be enabled to continue as mysqld was configured with --with-aria-tmp-tables
|
2019-07-08 1:39:51 139741663909696 [ERROR] Aborting
|
The test below mostly hits error 192, like in 90% cases; error 126 or error 22 (for which this report is devoted) are more rare.
Error 22 is currently best reproducible on 10.2, but it has been observed on all 10.2+ versions.
To run the test:
git clone https://github.com/MariaDB/randgen --branch mdev19980 rqg-mdev19980
|
cd rqg-mdev19980
|
. ./mdev19980.cmd <your basedir> <your vardir>
|
Important note: The command above will re-run the same test 10 times, regardless the outcome, and for each failure it will create <your vardir>_trialX directory, which you can later inspect. If you want it to stop on first failure, remove -force from the command line inside mdev19980.cmd, but keep in mind the note above, that error 22 represents the minority of failures.
I think that using on-disk vardir location gives better chances for reproducing (any of the errors), but it's unproved, I don't have statistics for this.