Uploaded image for project: 'MariaDB Server'
  1. MariaDB Server
  2. MDEV-32242

innodb.doublewrite test case always gets skipped

Details

    Description

      Re-write the `innodb.doublewrite` test case. Because it is being skipped everytime
      on debug build.

       
      innodb.doublewrite '32k,innodb,strict_crc32' [ skipped ]  Extra checkpoint 1 after 56238 (0:12=56344,0:11=56285)
      innodb.doublewrite '64k,innodb,strict_crc32' [ skipped ]  Extra checkpoint 1 after 55920 (0:10=56256,0:11=56299)
      innodb.doublewrite '32k,innodb,strict_full_crc32' [ skipped ]  Extra checkpoint 1 after 54403 (0:10=56269,0:9=56226)
      innodb.doublewrite '64k,innodb,strict_full_crc32' [ skipped ]  Extra checkpoint 1 after 55323 (0:10=56256,0:11=56299)
      innodb.doublewrite '16k,innodb,strict_crc32' [ skipped ]  Extra checkpoint 1 after 55488 (0:10=56410,0:11=56453)
      innodb.doublewrite '4k,innodb,strict_crc32' [ skipped ]  Extra checkpoint 1 after 74456 (0:12=74546,0:11=74503)
      innodb.doublewrite '8k,innodb,strict_crc32' [ skipped ]  Extra checkpoint 1 after 62190 (0:14=62280,0:13=62237)
      innodb.doublewrite '16k,innodb,strict_full_crc32' [ skipped ]  Extra checkpoint 1 after 56416 (0:12=56506,0:11=56463)
      innodb.doublewrite '4k,innodb,strict_full_crc32' [ skipped ]  Extra checkpoint 1 after 74497 (0:14=74587,0:13=74544)
      innodb.doublewrite '8k,innodb,strict_full_crc32' [ skipped ]  Extra checkpoint 1 after 60858 (0:12=62204,0:11=62161)
      

      InnoDB should avoid usage of innodb_buf_flush_list_now variable. It triggers the
      checkpoint very often and it leads to skip of this test case.

      Attachments

        Issue Links

          Activity

            Thank you. It is great to have this test back. I asked for some clarification, because I do not understand what happens. It would be better to replace the sleep 1 with some wait_condition.inc, to make the test faster and more reliable.

            marko Marko Mäkelä added a comment - Thank you. It is great to have this test back. I asked for some clarification, because I do not understand what happens. It would be better to replace the sleep 1 with some wait_condition.inc , to make the test faster and more reliable.

            I think that I found a way to execute the .ibd file specific part of the test in non-debug builds, by using FLUSH TABLES. But, my version of the test revealed some bugs in our recovery logic, which will have to be revised.

            marko Marko Mäkelä added a comment - I think that I found a way to execute the .ibd file specific part of the test in non-debug builds, by using FLUSH TABLES . But, my version of the test revealed some bugs in our recovery logic, which will have to be revised.

            People

              thiru Thirunarayanan Balathandayuthapani
              thiru Thirunarayanan Balathandayuthapani
              Votes:
              0 Vote for this issue
              Watchers:
              2 Start watching this issue

              Dates

                Created:
                Updated:
                Resolved:

                Git Integration

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