Uploaded image for project: 'MariaDB Server'
  1. MariaDB Server
  2. MDEV-39177

When tests are skipped, number of variations seem random

    XMLWordPrintable

Details

    • Bug
    • Status: In Testing (View Workflow)
    • Minor
    • Resolution: Unresolved
    • 11.8
    • 10.6
    • Tests, MTR
    • None

    Description

      While developing a new AppArmor profile for MariaDB I ran the mariadb-test-run with maximum number of tests trying to hit all code paths. Across runs I compared test counts to ensure the number of skipped tests kept going down as I fixed issues in the mtr feature detection code to get as many tests going as possible.

      While doing this I noticed that then mariadb-test-run skips tests, the number of variations is generates (and skips) seems random.

      As an example this test, when skipped, has two skipped variations:

      1)

      binlog_encryption.binlog_incident	skipped	Requires debug build
      

      2)

      binlog_encryption.binlog_incident 'mix'	skipped	Requires debug build
      binlog_encryption.binlog_incident 'row'	skipped	Requires debug build
      binlog_encryption.binlog_incident 'stmt'	skipped	Requires debug build
      

      I did not spot any pattern on why this happens, seems rather random. Here are the occurences in multiple test logs:

      $ zgrep --count -F binlog_encryption.binlog_incident *.gz
      20260312_164306_141e8.log.gz:3
      20260313_135410_5de3e.log.gz:3
      20260314_122209_ce752.log.gz:1
      20260315_100035_bb5f9.log.gz:3
      20260316_084907_eba9c.log.gz:3
      20260321_040838_43f11.log.gz:3
      20260321_041125_3f63e.log.gz:1
      

      There is no pattern when the variations appear. The same test run might have variations on some tests when skipped, but not others. There is no pattern on what the tests are. Below showing the randomness in two identical consecutive test runs on the same source code:

      $ zgrep -F binlog_encryption.mysqlbinlog 20260321_040838_43f11.log.gz 20260321_041125_3f63e.log.gz
      20260321_040838_43f11.log.gz:8806s binlog_encryption.mysqlbinlog            [ skipped ]  Requires debug build
      20260321_041125_3f63e.log.gz:8681s binlog_encryption.mysqlbinlog 'cmpr'     [ skipped ]  Requires debug build
      20260321_041125_3f63e.log.gz:8681s binlog_encryption.mysqlbinlog 'ncmpr'    [ skipped ]  Requires debug build
       
      $ zgrep -F binlog_encryption.binlog_incident 20260321_040838_43f11.log.gz 20260321_041125_3f63e.log.gz
      20260321_040838_43f11.log.gz:8806s binlog_encryption.binlog_incident 'mix'  [ skipped ]  Requires debug build
      20260321_040838_43f11.log.gz:8806s binlog_encryption.binlog_incident 'row'  [ skipped ]  Requires debug build
      20260321_040838_43f11.log.gz:8806s binlog_encryption.binlog_incident 'stmt' [ skipped ]  Requires debug build
      20260321_041125_3f63e.log.gz:8681s binlog_encryption.binlog_incident        [ skipped ]  Requires debug build
       
      $ zgrep -F innodb.auto_increment_dup 20260321_040838_43f11.log.gz 20260321_041125_3f63e.log.gz
      20260321_040838_43f11.log.gz:8806s innodb.auto_increment_dup                [ skipped ]  Requires debug build
      20260321_041125_3f63e.log.gz:8681s innodb.auto_increment_dup 'log-bin'      [ skipped ]  Requires debug build
      20260321_041125_3f63e.log.gz:8681s innodb.auto_increment_dup 'skip-log-bin' [ skipped ]  Requires debug build
       
      $ zgrep -F innodb.ddl_purge 20260321_040838_43f11.log.gz 20260321_041125_3f63e.log.gz
      20260321_040838_43f11.log.gz:8806s innodb.ddl_purge 'compact'               [ skipped ]  Requires debug build
      20260321_040838_43f11.log.gz:8806s innodb.ddl_purge 'dynamic'               [ skipped ]  Requires debug build
      20260321_040838_43f11.log.gz:8806s innodb.ddl_purge 'redundant'             [ skipped ]  Requires debug build
      20260321_041125_3f63e.log.gz:8681s innodb.ddl_purge                         [ skipped ]  Requires debug build
      

      I tried to come up with a patch to fix this myself, but I didn't figure out what triggers the variations by reading the code or by reading AI generated docs about mariadb-test-run or mysqltest execution, hence filing now this bug report to see if others have noticed this bug, and if there are ideas on how to fix it.

      Attachments

        Activity

          People

            serg Sergei Golubchik
            otto Otto Kekäläinen
            Votes:
            0 Vote for this issue
            Watchers:
            2 Start watching this issue

            Dates

              Created:
              Updated:

              Git Integration

                Error rendering 'com.xiplink.jira.git.jira_git_plugin:git-issue-webpanel'. Please contact your Jira administrators.