Uploaded image for project: 'Lustre'
  1. Lustre
  2. LU-3970

Add procfs interface for clearing lustre caches in parallel

Details

    • Improvement
    • Resolution: Won't Fix
    • Minor
    • None
    • None

    Description

      Cray experienced slowness clearing the lustre caches after job termination. This slowness is a result of clearing the caches for each namespace in series. Lustre should provide a high-level procfs interface which clears lustre caches across namespaces in parallel and returns when all caches are cleared.

      The interface should be at /proc/fs/lustre/ldlm/drop_caches, and when written to, it must clear all lustre caches in parallel and return when finished.

      It must have the same effect as the following, but in parallel:

      for LRU in $(ls /proc/fs/lustre/ldlm/namespaces/*osc*/lru_size); do
    echo clear > $LRU
done

for LRU in $(ls /proc/fs/lustre/ldlm/namespaces/*mdc*/lru_size); do
    echo clear > $LRU
done

      Attachments

        Issue Links

          is related to

          Improvement - An improvement or enhancement to an existing feature or task. LU-5134 Add option to lctl set_param for setting parameters in parallel

          • Minor - Minor loss of function, or other problem where easy workaround is present.
          • Resolved

          Activity

            [LU-3970] Add procfs interface for clearing lustre caches in parallel
            jhammond John Hammond made changes -
            Description Original: Cray experienced slowness clearing the lustre caches after job termination. This slowness is a result of clearing the caches for each namespace in series. Lustre should provide a high-level procfs interface which clears lustre caches across namespaces in parallel and returns when all caches are cleared.

            The interface should be at /proc/fs/lustre/ldlm/drop_caches, and when written to, it must clear all lustre caches in parallel and return when finished.

            It must have the same effect as the following, but in parallel:

            for LRU in $(ls /proc/fs/lustre/ldlm/namespaces/*osc*/lru_size); do
                echo clear > $LRU
            done

            for LRU in $(ls /proc/fs/lustre/ldlm/namespaces/*mdc*/lru_size); do
                echo clear > $LRU
            done             		New: Cray experienced slowness clearing the lustre caches after job termination. This slowness is a result of clearing the caches for each namespace in series. Lustre should provide a high-level procfs interface which clears lustre caches across namespaces in parallel and returns when all caches are cleared.

            The interface should be at /proc/fs/lustre/ldlm/drop_caches, and when written to, it must clear all lustre caches in parallel and return when finished.

            It must have the same effect as the following, but in parallel:
            {noformat}
            for LRU in $(ls /proc/fs/lustre/ldlm/namespaces/*osc*/lru_size); do
                echo clear > $LRU
            done

            for LRU in $(ls /proc/fs/lustre/ldlm/namespaces/*mdc*/lru_size); do
                echo clear > $LRU
            done
            {noformat}
            pjones Peter Jones made changes -
            Resolution New: Won't Fix [ 2 ]
            Status Original: Open [ 1 ] New: Resolved [ 5 ]
            pjones Peter Jones made changes -
            Link New: This issue is related to LU-5134 [ LU-5134 ]
            yujian Jian Yu made changes -
            Assignee Original: WC Triage [ wc-triage ] New: Jian Yu [ yujian ]
            pjones Peter Jones made changes -
            Labels Original: performance New: patch performance
            haasken Ryan Haasken created issue -

            People

              yujian Jian Yu
              haasken Ryan Haasken
              Votes:
              0 Vote for this issue
              Watchers:
              6 Start watching this issue

              Dates

                Created:
                Updated:
                Resolved: