Details

    • Type: Task
    • Status: Closed
    • Resolution: Fixed
    • Affects Version/s: None
    • Fix Version/s: WT2.5.2
    • Labels:

      Description

      I was looking at some statistics via the wtstats graph tool, and they don't appear to match those in the original file.

      The two statistics are:

      tracked dirty bytes in the cache
      

      and

      bytes currently in the cache
      

      The wtstats graph I see for those two statistics is:

      An extract of the raw data is:

      Mar 10 06:00:07 13937645434 /mnt/fast/alexg/data cache: tracked dirty bytes in the cache
      Mar 10 06:00:07 15032413045 /mnt/fast/alexg/data cache: bytes currently in the cache
      --
      Mar 10 06:00:17 14118949794 /mnt/fast/alexg/data cache: tracked dirty bytes in the cache
      Mar 10 06:00:17 15031434886 /mnt/fast/alexg/data cache: bytes currently in the cache
      --
      Mar 10 06:00:27 14245210823 /mnt/fast/alexg/data cache: tracked dirty bytes in the cache
      Mar 10 06:00:27 15032327122 /mnt/fast/alexg/data cache: bytes currently in the cache
      --
      Mar 10 06:00:37 13829546010 /mnt/fast/alexg/data cache: tracked dirty bytes in the cache
      Mar 10 06:00:37 14839582962 /mnt/fast/alexg/data cache: bytes currently in the cache
      --
      Mar 10 06:00:47 13598968927 /mnt/fast/alexg/data cache: tracked dirty bytes in the cache
      Mar 10 06:00:47 15030610981 /mnt/fast/alexg/data cache: bytes currently in the cache
      --
      Mar 10 06:00:57 13407003442 /mnt/fast/alexg/data cache: tracked dirty bytes in the cache
      Mar 10 06:00:57 15032283870 /mnt/fast/alexg/data cache: bytes currently in the cache
      --
      Mar 10 06:01:07 14325052454 /mnt/fast/alexg/data cache: tracked dirty bytes in the cache
      Mar 10 06:01:07 15031068774 /mnt/fast/alexg/data cache: bytes currently in the cache
      --
      Mar 10 06:01:17 14373973388 /mnt/fast/alexg/data cache: tracked dirty bytes in the cache
      Mar 10 06:01:17 15031787034 /mnt/fast/alexg/data cache: bytes currently in the cache
      --
      Mar 10 06:01:27 14959868383 /mnt/fast/alexg/data cache: tracked dirty bytes in the cache
      Mar 10 06:01:27 15032397299 /mnt/fast/alexg/data cache: bytes currently in the cache
      --
      Mar 10 06:01:37 14865619787 /mnt/fast/alexg/data cache: tracked dirty bytes in the cache
      Mar 10 06:01:37 15032398484 /mnt/fast/alexg/data cache: bytes currently in the cache
      

      I wonder whether the tool is treating the tracked dirty bytes as a running total rather than an absolute number?

        Issue Links

          Activity

          Hide
          rueckstiess Thomas Rückstieß (Inactive) added a comment -

          I think this is because "bytes currently in the cache" is in the no_clear_list, but "tracked dirty bytes in the cache" is not. See here.

          If a stat is not on this list, wtstats.py will compute the difference to the last value and plot that (code).

          There is also a --clear option to influence the behavior of wtstats.py. If this option is set, all values are treated as absolute and no difference is computed.

          The question is: should the "bytes currently in the cache" stat also be added to that list? Are there others that need to be added?

          Show
          rueckstiess Thomas Rückstieß (Inactive) added a comment - I think this is because "bytes currently in the cache" is in the no_clear_list , but "tracked dirty bytes in the cache" is not. See here . If a stat is not on this list, wtstats.py will compute the difference to the last value and plot that ( code ). There is also a --clear option to influence the behavior of wtstats.py. If this option is set, all values are treated as absolute and no difference is computed. The question is: should the "bytes currently in the cache" stat also be added to that list? Are there others that need to be added?
          Hide
          agorrod Alex Gorrod added a comment -

          Thanks Thomas Rückstieß I've pushed a fix for the dirty bytes change and one other.

          [~keithbostic] Could you do a pass over the flags for statistics values as well? If you are happy please close the issue.

          Show
          agorrod Alex Gorrod added a comment - Thanks Thomas Rückstieß I've pushed a fix for the dirty bytes change and one other. [~keithbostic] Could you do a pass over the flags for statistics values as well? If you are happy please close the issue.
          Hide
          agorrod Alex Gorrod added a comment -

          [~keithbostic] Would you mind doing a pass over stat_data.py and checking to see whether all the flags values make sense to you in light of this issue?

          Show
          agorrod Alex Gorrod added a comment - [~keithbostic] Would you mind doing a pass over stat_data.py and checking to see whether all the flags values make sense to you in light of this issue?
          Hide
          keith.bostic Keith Bostic added a comment -

          @agorrod, I did a pass and made some changes – can you please review 4790f13, and then I think WT-1742 is ready to close.

          Show
          keith.bostic Keith Bostic added a comment - @agorrod, I did a pass and made some changes – can you please review 4790f13, and then I think WT-1742 is ready to close.
          Hide
          agorrod Alex Gorrod added a comment -

          Thanks Thomas Rückstieß lgtm. This issue is done for now.

          Show
          agorrod Alex Gorrod added a comment - Thanks Thomas Rückstieß lgtm. This issue is done for now.
          Hide
          ramon.fernandez Ramon Fernandez added a comment -

          Additional ticket information from GitHub

          This ticket was referenced in the following commits:
          Show
          ramon.fernandez Ramon Fernandez added a comment - Additional ticket information from GitHub This ticket was referenced in the following commits: 89db28287d7b212efb31203eca253fbf144ea207 4790f13cd8191d834f40c97a0c00d729f91acd1b

            People

            • Votes:
              0 Vote for this issue
              Watchers:
              1 Start watching this issue

              Dates

              • Created:
                Updated:
                Resolved:
                Days since reply:
                2 years, 1 week, 5 days ago
                Date of 1st Reply: