Uploaded image for project: 'MariaDB Server'
  1. MariaDB Server
  2. MDEV-11027

InnoDB log recovery is too noisy

Details

    • 5.5.55, 10.2.5-1

    Attachments

      1. jan.opt
        0.1 kB
      2. jan.test
        0.9 kB

      Issue Links

        causes

        Bug - A problem which impairs or prevents the functions of the product. MDEV-13814 Extra logging when innodb_log_archive=ON ( [Note] InnoDB: Read redo log up to LSN=...)

        • Major - Major loss of function.
        • Closed

        Bug - A problem which impairs or prevents the functions of the product. MDEV-14174 crash on start with innodb-track-changed-pages

        • Major - Major loss of function.
        • Closed
        relates to

        Bug - A problem which impairs or prevents the functions of the product. MDEV-6456 No progress indication for "Reading tablespace information from the .ibd files"

        • Major - Major loss of function.
        • Closed

        Bug - A problem which impairs or prevents the functions of the product. MDEV-10509 Excessive InnoDB logging on release server

        • Major - Major loss of function.
        • Closed

        Bug - A problem which impairs or prevents the functions of the product. MDEV-12103 Reduce the time of looking for MLOG_CHECKPOINT during crash recovery

        • Major - Major loss of function.
        • Closed

        Task - A task that needs to be done. MDEV-14705 systemd: EXTEND_TIMEOUT_USEC= to avoid startup and shutdown timeouts

        • Critical - Crashes, loss of data, severe memory leak.
        • Closed

        Bug - A problem which impairs or prevents the functions of the product. MDEV-29911 InnoDB recovery and mariadb-backup --prepare fail to report detailed progress

        • Critical - Crashes, loss of data, severe memory leak.
        • Closed

        Activity

          Ascending order - Click to sort in descending order
          View 9 older comments
          marko Marko Mäkelä added a comment -

          sjmudd, would you be happy with rate-limited output like this?

          2017-03-07 14:58:57 4147463936 [Note] InnoDB: Starting crash recovery from checkpoint LSN=1238777
          		 
          2017-03-07 14:58:57 4147463936 [Note] InnoDB: Read redo log up to LSN=1242112
          		 
          2017-03-07 14:58:57 4147463936 [Note] InnoDB: Starting log recovery batch of 12 pages
          		 
          2017-03-07 14:58:57 4147463936 [Note] InnoDB: To recover: 12 pages from log

          The first and third message would be output unconditionally, while the second and last message would only be output at most once per 15 seconds. The second message would be issued directly after completing redo log file I/O, while the last message would be issued at the end of recv_recover_page(), which is indirectly tied to data file I/O.

          It would be tricky to add progress reporting to the page writes, so I would avoid it in the first cut.
          On a related note, MDEV-10509 (another major source of InnoDB spam in MariaDB) should be addressed by implementing MariaDB progress reporting for ALTER TABLE…ALGORITHM=INPLACE.

          marko Marko Mäkelä added a comment - sjmudd , would you be happy with rate-limited output like this? 2017-03-07 14:58:57 4147463936 [Note] InnoDB: Starting crash recovery from checkpoint LSN=1238777 2017-03-07 14:58:57 4147463936 [Note] InnoDB: Read redo log up to LSN=1242112 2017-03-07 14:58:57 4147463936 [Note] InnoDB: Starting log recovery batch of 12 pages 2017-03-07 14:58:57 4147463936 [Note] InnoDB: To recover: 12 pages from log The first and third message would be output unconditionally, while the second and last message would only be output at most once per 15 seconds. The second message would be issued directly after completing redo log file I/O, while the last message would be issued at the end of recv_recover_page(), which is indirectly tied to data file I/O. It would be tricky to add progress reporting to the page writes, so I would avoid it in the first cut. On a related note, MDEV-10509 (another major source of InnoDB spam in MariaDB) should be addressed by implementing MariaDB progress reporting for ALTER TABLE…ALGORITHM=INPLACE.
          marko Marko Mäkelä added a comment - - edited

          bb-10.1-marko
          bb-10.2-marko
          I would also appreciate some feedback from users.

          marko Marko Mäkelä added a comment - - edited bb-10.1-marko bb-10.2-marko I would also appreciate some feedback from users.
          jplindst Jan Lindström (Inactive) added a comment -

          ok to push, did you consider 10.0/5.5 ?

          jplindst Jan Lindström (Inactive) added a comment - ok to push, did you consider 10.0/5.5 ?
          marko Marko Mäkelä added a comment -

          Thanks for the review! I will backport; the risk should be very low.
          The patches will be a bit different, because InnoDB in 5.5 is in C, not in C++. Also the sd_notifyf() calls for systemd integration are not present before 10.1.

          marko Marko Mäkelä added a comment - Thanks for the review! I will backport; the risk should be very low. The patches will be a bit different, because InnoDB in 5.5 is in C, not in C++. Also the sd_notifyf() calls for systemd integration are not present before 10.1.
          marko Marko Mäkelä added a comment -

          Finally, the merge from 10.0 to 10.1 is completed as well.

          marko Marko Mäkelä added a comment - Finally, the merge from 10.0 to 10.1 is completed as well.

          People

            marko Marko Mäkelä
            jplindst Jan Lindström (Inactive)
            Votes:
            1 Vote for this issue
            Watchers:
            4 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.