Details

    • Type: Bug
    • Status: Resolved
    • Priority: Major
    • Resolution: Duplicate
    • Affects Version/s: None
    • Fix Version/s: None
    • Component/s: None
    • Labels:
      None

      Description

      I was running RandomWalk and ran into an issue where one of my walkers got stuck doing a merge (on table id=20). The only related error that showed up was that a file in HDFS couldn't be found (and continues to not be found, it keeps checking regularly) while trying to merge the metadata records. Here's the error:

      2013-08-14 11:54:27,246 [master.Master] ERROR: Unable merge metadata table records
      org.apache.accumulo.core.client.AccumuloException: java.io.IOException: Could not find file /20//c-00000000 in [hdfs://localhost:9000/accumulo/tables]
              at org.apache.accumulo.server.master.TabletGroupWatcher.mergeMetadataRecords(TabletGroupWatcher.java:548)
              at org.apache.accumulo.server.master.TabletGroupWatcher.updateMergeState(TabletGroupWatcher.java:368)
              at org.apache.accumulo.server.master.TabletGroupWatcher.run(TabletGroupWatcher.java:263)
      Caused by: java.io.IOException: Could not find file /20//c-00000000 in [hdfs://localhost:9000/accumulo/tables]
              at org.apache.accumulo.server.fs.VolumeManagerImpl.getFullPath(VolumeManagerImpl.java:446)
              at org.apache.accumulo.server.util.MetadataTableUtil.createDeleteMutation(MetadataTableUtil.java:501)
              at org.apache.accumulo.server.master.TabletGroupWatcher.mergeMetadataRecords(TabletGroupWatcher.java:502)
              ... 2 more
      

      And the metadata table had c-00000000 listed here:

      root@802 !METADATA> grep c-00000
      20;00b581d7479230dd srv:dir []    /c-00000000
      20< srv:dir []    /c-00000001
      

      I was testing my table namespaces code but this seemed like an unrelated issue.

        Issue Links

          Activity

          Hide
          vines John Vines added a comment -

          Version?

          Show
          vines John Vines added a comment - Version?
          Hide
          shickey Sean Hickey added a comment -

          Sorry, 1.6, trunk/master, whatever you want to call it.

          Show
          shickey Sean Hickey added a comment - Sorry, 1.6, trunk/master, whatever you want to call it.
          Hide
          vines John Vines added a comment -

          This is the same issue I saw in 1745, which got fixed by SOMETHING else.

          Show
          vines John Vines added a comment - This is the same issue I saw in 1745, which got fixed by SOMETHING else.
          Hide
          samiunn M Sami added a comment -

          I'm sorry to resurrect this, but I'm having the same problem and it's flooded the logs interrmittenly. It's a cloudera's accumulo parcel. I wonder if it's the right place for me to comment here?

          Show
          samiunn M Sami added a comment - I'm sorry to resurrect this, but I'm having the same problem and it's flooded the logs interrmittenly. It's a cloudera's accumulo parcel. I wonder if it's the right place for me to comment here?
          Hide
          elserj Josh Elser added a comment -

          Hi M Sami, you should contact Cloudera for support. If you are running into this issue, it sounds like they haven't patched ACCUMULO-1745 which is the real issue.

          Show
          elserj Josh Elser added a comment - Hi M Sami , you should contact Cloudera for support. If you are running into this issue, it sounds like they haven't patched ACCUMULO-1745 which is the real issue.
          Hide
          elserj Josh Elser added a comment -

          Actually, re-reading through the linked issues, there was no code change for a defect. The only thing added were tests. It's possible there's another issue that happened to fix a bug, but I'm not sure. We're happy to help you on the mailing lists with Apache Accumulo defects, but vendor specific issues are better served on that vendor's support channels.

          Show
          elserj Josh Elser added a comment - Actually, re-reading through the linked issues, there was no code change for a defect. The only thing added were tests. It's possible there's another issue that happened to fix a bug, but I'm not sure. We're happy to help you on the mailing lists with Apache Accumulo defects, but vendor specific issues are better served on that vendor's support channels.
          Hide
          samiunn M Sami added a comment -

          Thank you for the suggestion, I really appreciate it.

          Show
          samiunn M Sami added a comment - Thank you for the suggestion, I really appreciate it.

            People

            • Assignee:
              Unassigned
              Reporter:
              shickey Sean Hickey
            • Votes:
              0 Vote for this issue
              Watchers:
              3 Start watching this issue

              Dates

              • Created:
                Updated:
                Resolved:

                Development