PS-5541: Do not try to decrypt blocks in changed page tracking. (5.7) #3150
Conversation
dutow
changed the title
mysql-test/t/percona_bug_ps5541.test
Outdated
| @@ -0,0 +1,47 @@ | |||
| -- source include/big_test.inc | |||
There was a problem hiding this comment.
If I understand correctly, the test does not require redo log encryption itself to be enabled, correct? In which case this scenario should be test by existing innodb.percona_changed* testcases, please check, whether the new testcase is required at all
There was a problem hiding this comment.
That's right, encryption is not needed - but a huge amount of data is. If I insert fewer records than what we have in this test case, the issue won't show up without this commit. And with this amount, it runs for 15-20 minutes on a debug build.
There was a problem hiding this comment.
OK, then the existing bitmap testcases have predetermined name prefix (percona_changed_page_bmp in general, percona_changes_pages for I_S queries like this one), so please rename to percona_changed_pages_big and add a header comment --echo # / --echo # PS-5541: blah blah / --echo #
mysql-test/t/percona_bug_ps5541.test
Outdated
| @@ -0,0 +1,47 @@ | |||
| -- source include/big_test.inc | |||
There was a problem hiding this comment.
OK, then the existing bitmap testcases have predetermined name prefix (percona_changed_page_bmp in general, percona_changes_pages for I_S queries like this one), so please rename to percona_changed_pages_big and add a header comment --echo # / --echo # PS-5541: blah blah / --echo #
mysql-test/t/percona_bug_ps5541.test
Outdated
| show status like 'Innodb_page%'; | ||
| --enable_result_log | ||
|
|
||
| USE test; |
mysql-test/t/percona_bug_ps5541.test
Outdated
| -- source include/big_test.inc | ||
|
|
||
| --disable_result_log | ||
| show status like 'Innodb_page%'; |
mysql-test/t/percona_bug_ps5541.test
Outdated
| @@ -0,0 +1,47 @@ | |||
| -- source include/big_test.inc | |||
|
|
|||
There was a problem hiding this comment.
--source include/have_innodb.inc
mysql-test/t/percona_bug_ps5541.test
Outdated
| `g` int(11) NOT NULL, | ||
| PRIMARY KEY (`i`), | ||
| KEY key_g (`g`) | ||
| ) ENGINE=InnoDB DEFAULT CHARSET=latin1; |
There was a problem hiding this comment.
| ) ENGINE=InnoDB DEFAULT CHARSET=latin1; | |
| ) ENGINE=InnoDB DEFAULT CHARSET=latin1; |
mysql-test/t/percona_bug_ps5541.test
Outdated
| KEY key_g (`g`) | ||
| ) ENGINE=InnoDB DEFAULT CHARSET=latin1; | ||
|
|
||
| INSERT INTO joinit VALUES (NULL, uuid(), time(now()), (FLOOR( 1 + RAND( ) *60 ))); |
There was a problem hiding this comment.
Everywhere
| INSERT INTO joinit VALUES (NULL, uuid(), time(now()), (FLOOR( 1 + RAND( ) *60 ))); | |
| INSERT INTO joinit VALUES (NULL, uuid(), time(now()), (FLOOR(1 + RAND() * 60))); |
mysql-test/t/percona_bug_ps5541.test
Outdated
| INSERT INTO joinit SELECT NULL, uuid(), time(now()), (FLOOR( 1 + RAND( ) *60 )) FROM joinit; | ||
|
|
||
| --disable_result_log | ||
| show status like 'Innodb_page%'; |
| @@ -0,0 +1,2 @@ | |||
| --innodb_track_changed_pages=1 --innodb_max_bitmap_file_size=1G --innodb_buffer_pool_size=1G | |||
There was a problem hiding this comment.
Is 1G buffer pool size necessary?
| KEY key_g (`g`) | ||
| ) ENGINE=InnoDB DEFAULT CHARSET=latin1; | ||
|
|
||
| INSERT INTO joinit VALUES (NULL, uuid(), time(now()), FLOOR(1 + RAND() *60 )); |
There was a problem hiding this comment.
Comment not addressed
| INSERT INTO joinit SELECT NULL, uuid(), time(now()), FLOOR(1 + RAND() *60) FROM joinit; | ||
|
|
||
| --disable_result_log | ||
| select count(*) as count_innodb_changed_pages from information_schema.INNODB_CHANGED_PAGES; |
There was a problem hiding this comment.
Make SQL case consistent
| KEY key_g (`g`) | ||
| ) ENGINE=InnoDB DEFAULT CHARSET=latin1; | ||
|
|
||
| INSERT INTO joinit VALUES (NULL, uuid(), time(now()), FLOOR(1 + RAND() * 60)); |
There was a problem hiding this comment.
Strictly speaking
| INSERT INTO joinit VALUES (NULL, uuid(), time(now()), FLOOR(1 + RAND() * 60)); | |
| INSERT INTO joinit VALUES (NULL, uuid(), time(now()), FLOOR(1 + RAND() * 60)); |
still not done but of course not enough to cause a PR respin by itself
Issue: changed page tracking uses the LOG flag during read operations to follow the correct code path. Redo log encryption backported from 8.0 tries to decrypt pages with a certain bit set, and fails. While the end result is the same, the error log gets flooded with decryption errors. Solution: the NO_ENCRYPTION flag, previously ignored by log encryption now disables decryption during log reads too.
Issue: changed page tracking uses the LOG flag during read
operations to follow the correct code path. Redo log encryption
backported from 8.0 tries to decrypt pages with a certain bit set,
and fails. While the end result is the same, the error log gets
flooded with decryption errors.
Solution: the NO_ENCRYPTION flag, previously ignored by log
encryption now disables decryption during log reads too.