Details
-
Bug
-
Status: Closed (View Workflow)
-
Major
-
Resolution: Fixed
-
10.2(EOL), 10.3(EOL), 10.4(EOL)
-
None
Description
http://buildbot.askmonty.org/buildbot/builders/kvm-fulltest2/builds/9687
encryption.innodb-redo-badkey 'cbc,innodb' w2 [ fail ] Found warnings/errors in server log file!
|
Test ended at 2017-09-20 15:48:54
|
line
|
2017-09-20 15:48:48 3072805184 [ERROR] InnoDB: The page [page id: space=4, page number=1] in file './test/t1.ibd' cannot be decrypted.
|
2017-09-20 15:48:48 2939935552 [ERROR] InnoDB: The page [page id: space=4, page number=2] in file './test/t1.ibd' cannot be decrypted.
|
2017-09-20 15:48:49 2939935552 [Warning] InnoDB: Table in tablespace 4 encrypted. However key management plugin or used key_version 0is not found or used encryption algorithm or method does not match. Can't continue opening the table.
|
^ Found warnings in /mnt/buildbot/build/mariadb-10.2.9/mysql-test/var/2/log/mysqld.1.err
|
ok
|
|
|
- saving '/mnt/buildbot/build/mariadb-10.2.9/mysql-test/var/2/log/encryption.innodb-redo-badkey-cbc,innodb/' to '/mnt/buildbot/build/mariadb-10.2.9/mysql-test/var/log/encryption.innodb-redo-badkey-cbc,innodb/'
|
Attachments
Issue Links
- relates to
-
-
- Closed
-
-
-
- Closed
-
Activity
I still don't understand the purpose of this record, be it a warning or a note. If InnoDB already knows that the primary copy is fine, then why check the redundant one, and if it does need to check on whatever reason, why write about it?
Or, if a redundant one is checked first on whatever reason and found corrupt, InnoDB still doesn't know if it's even needed, so why not to suppress the message until it actually knows whether it was needed, and then it should be either an error (since it's needed but corrupt), or shouldn't be written at all, since nobody cares because the primary one is fine.
I think that we should remove this message already in 10.1, even though 10.1 will be more likely to suffer from actual recovery failures due to MDEV-12699.
The originally reported message (in the bug title) mostly went away thanks to MDEV-12699, which was not fixed in the 10.1 series. As part of this ticket, we only removed a warning for copies of pages that are corrupted in the doublewrite buffer.
Even though
MDEV-12699fixed most of this in 10.2+, I think that we should fix this failure by changing the warning message to a note:10.1 5b035c84567b60c2ca680de96f5179a6
encryption.innodb-redo-badkey 'ctr,innodb_plugin' w1 [ fail ] Found warnings/errors in server log file!Test ended at 2019-04-30 08:45:25line2019-04-30 8:45:17 3068614656 [Warning] InnoDB: A doublewrite copy of page 4:1 is corrupted.^ Found warnings in /mnt/buildbot/build/mariadb-10.1.39/mysql-test/var/1/log/mysqld.1.errokFor the primary copy of the page, a warning may be fine. But, I do not think that we should issue warnings regarding doublewrite buffer corruption.