Uploaded image for project: 'Solr'
  1. Solr
  2. SOLR-15998

replicaType=TLOG oom killer event occurs

Attach filesAttach ScreenshotAdd voteVotersWatch issueWatchersCreate sub-taskLinkCloneUpdate Comment AuthorReplace String in CommentUpdate Comment VisibilityDelete Comments
    XMLWordPrintableJSON

Details

    • Improvement
    • Status: Open
    • Critical
    • Resolution: Unresolved
    • 8.4.1
    • None
    • clients - java
    • None

    Description

      Hello,

      This is my Solr Cloud cluster.

       

      16 Shard (32 Node)
          1 LEADER   (replicaType=TLOG)
          1 REPLICA  (replicaType=TLOG)

      Node Information
          4core vCPU
          32 GB Pysical Memory
          16~20GB SOLR - JAVA HEAP
          500 GB SSD

      Traffic
          - Write 10,000 ~ 30,000 document/Per sec  

          - Solr Index Size - using 50% disk 

      In this situation, all replicas are down when performing Select Query. (Solr_oom.sh killer event occurs)

      However, this does not happen in case of replicaType=NRT.

      But, NRT write performance is severely degraded, so we try not to use it.

      Using cursor. 

       

      Attachments

        Activity

          This comment will be Viewable by All Users Viewable by All Users
          Cancel

          People

            Unassigned Unassigned
            rijachung28 Mark

            Dates

              Created:
              Updated:

              Slack

                Issue deployment