| Bug #87303 | innodb_redo_log_encrypt-setting testcases are unstable | ||
|---|---|---|---|
| Submitted: | 3 Aug 2017 13:01 | Modified: | 9 Aug 2017 13:59 |
| Reporter: | Laurynas Biveinis (OCA) | Email Updates: | |
| Status: | Closed | Impact on me: | |
| Category: | MySQL Server: Tests | Severity: | S3 (Non-critical) |
| Version: | 8.0.2 | OS: | Any |
| Assigned to: | CPU Architecture: | Any | |
| Tags: | Contribution, innodb, mtr | ||
[4 Aug 2017 2:45]
Laurynas Biveinis
Exact same issue on innodb.log_encrypt_3:
@@ -6,7 +6,7 @@
Level Code Message
SELECT @@global.innodb_redo_log_encrypt ;
@@global.innodb_redo_log_encrypt
-0
+1
CREATE TABLE tde_db.t4 (a BIGINT PRIMARY KEY, b LONGBLOB) ENGINE=InnoDB;
INSERT INTO t4 (a, b) VALUES (1, REPEAT('a', 6*512*512));
SELECT a,LEFT(b,10) FROM tde_db.t4;
mysqltest: Result content mismatch
Caused by exact same antipattern:
SET GLOBAL innodb_redo_log_encrypt = 1;
--sleep 1
SHOW WARNINGS;
SELECT @@global.innodb_redo_log_encrypt ;
[4 Aug 2017 2:58]
Laurynas Biveinis
By code review innodb.log_encrypt_kill may fail to find an error message in the error log because of the same: SELECT @@global.innodb_redo_log_encrypt ; SET GLOBAL innodb_redo_log_encrypt = 1; SET GLOBAL innodb_undo_log_encrypt = 1; --sleep 1 # sleep to get message recorded in error log --sleep 1
[5 Aug 2017 7:34]
Laurynas Biveinis
Bug 87303 fix for 8.0.2 (*) I confirm the code being submitted is offered under the terms of the OCA, and that I am authorized to contribute it.
Contribution: bug87303-8.0.2.patch (application/octet-stream, text), 4.42 KiB.
[7 Aug 2017 7:38]
MySQL Verification Team
Hello Laurynas, Thank you for the report and contribution. Thanks, Umesh
[9 Aug 2017 13:59]
Paul DuBois
Posted by developer: Fixed in 8.0.3. The sys_vars.innodb_redo_log_encrypt_basic test case output was unstable. Thanks to Laurynas Biveinis for the patch.
Description: I saw only a single instance of this and was not able to repeat it, but the cause seems clear from the testcase. sys_vars.innodb_redo_log_encrypt_basic w4 [ fail ] Test ended at 2017-08-02 18:16:12 CURRENT_TEST: sys_vars.innodb_redo_log_encrypt_basic --- /mnt/workspace/mysql-5.7-param/BUILD_TYPE/release/Host/ubuntu-xenial-64bit/mysql-test/suite/sys_vars/r/innodb_redo_log_encrypt_basic.result 2017-08-02 18:21:30.204890246 +0300 +++ /mnt/workspace/mysql-5.7-param/BUILD_TYPE/release/Host/ubuntu-xenial-64bit/build/mysql-test/var/4/log/innodb_redo_log_encrypt_basic.reject 2017-08-02 21:16:12.355291397 +0300 @@ -25,7 +25,7 @@ set global innodb_redo_log_encrypt=1; select @@global.innodb_redo_log_encrypt; @@global.innodb_redo_log_encrypt -0 +1 select * from performance_schema.global_variables where variable_name='innodb_redo_log_encrypt'; VARIABLE_NAME VARIABLE_VALUE innodb_redo_log_encrypt OFF mysqltest: Result content mismatch How to repeat: Was not able to repeat. But see below Suggested fix: The testcase bit in question is set global innodb_redo_log_encrypt=1; --sleep 2 select @@global.innodb_redo_log_encrypt; where the select returns zero. It is the InnoDB master thread which reset the variable from one to zero by calling log_enable_encryption_if_set. The testcase uses sleep instead of proper sync to give the master thread a chance to run, but this is not guaranteed to happen on a loaded system. The sleep and select should be replaced by a wait until the variable value becomes zero.