Bug #113738 Test innodb.resurrection_logs failing if built with sanitizer
Submitted: 24 Jan 10:18 Modified: 24 Jan 12:14
Reporter: Laurynas Biveinis (OCA) Email Updates:
Status: Verified Impact on me:
None 
Category:MySQL Server: Tests Severity:S3 (Non-critical)
Version:8.3 OS:MacOS (14.3)
Assigned to: CPU Architecture:ARM

[24 Jan 10:18] Laurynas Biveinis
Description:
If server is built with sanitizers, the test failed with a result difference every time I tried. If not built with sanitizers, the test passes most of the time but not always. I suspect the cause is not the sanitizer itself, but that it adds enough slowness to expose some underlying instability.

$ mtr innodb.resurrection_logs
...
[ 50%] innodb.resurrection_logs                  [ fail ]
        Test ended at 2024-01-24 12:16:08

CURRENT_TEST: innodb.resurrection_logs
WARNING: MYSQL_OPT_RECONNECT is deprecated and will be removed in a future version.
WARNING: MYSQL_OPT_RECONNECT is deprecated and will be removed in a future version.
--- /Users/laurynas/vilniusdb/mysql-8.3.0/mysql-test/suite/innodb/r/resurrection_logs.result	2024-01-22 16:39:22
+++ /Users/laurynas/vilniusdb/mysql-8.3.0/_build-debug-san/mysql-test/var/log/resurrection_logs.reject	2024-01-24 13:16:08
@@ -31,7 +31,7 @@
 CALL bulk_upd(10);
 # Kill and restart: --log-error-verbosity=3 --debug=+d,resurrect_logs
 Pattern "Transaction ID:.* found for resurrecting inserts" found
-Pattern "Records read:.* - Pages read:.*" not found
+Pattern "Records read:.* - Pages read:.*" found
 Pattern "Transaction ID:.* found for resurrecting updates" found
 Pattern "Identified table ID:.* to acquire lock" found
 Pattern "Resurrected .* transaction.* doing inserts." found

mysqltest: Result content mismatch

How to repeat:
Server built with -DCMAKE_EXPORT_COMPILE_COMMANDS=ON -DCMAKE_BUILD_TYPE=Debug -DWITH_DEBUG=ON -DMYSQL_MAINTAINER_MODE=ON -DWITH_SYSTEM_LIBS=ON -DFORCE_COLORED_OUTPUT=ON -DWITH_ZLIB=bundled -DWITH_ASAN=ON -DWITH_ASAN_SCOPE=ON -DWITH_UBSAN=ON
[24 Jan 12:14] MySQL Verification Team
Hi Mr. Bibeinis,

Thank you for your bug report.

First of all, it took us some time to upgrade our macOS on M1 to the latest 14.3 version.

Then we ran the existing sanitizer built binaries and we managed to repeat the failure of the same test, after several attempts.

This is now a verified bug for the version 8.3 on the specific platform.