HBase
  1. HBase
  2. HBASE-3502

Can't open region because can't open .regioninfo because AlreadyBeingCreatedException

    Details

    • Type: Bug Bug
    • Status: Resolved
    • Priority: Critical Critical
    • Resolution: Fixed
    • Affects Version/s: None
    • Fix Version/s: 0.90.1
    • Component/s: None
    • Labels:
      None
    • Hadoop Flags:
      Reviewed

      Description

      Testing killing .META. I tripped over this one. Last thing seen on regionserver killed was:

      2011-02-02 21:44:48,379 DEBUG org.apache.hadoop.hbase.regionserver.HRegion: Instantiated TestTable,0591556500,1296683085472.76c9a32c5f068d16240e42a15fed8417.
      

      ... which means we could have been inside checkRegioninfoOnFilesystem when we were killed.

      This tries to create the .regioninfo file. Seems like that was started over at the NN but then the RS died shortly afterward. Its stopping the Region opening. I suppose I could try and open it for append to shut it then reopen?

        Activity

        Hide
        Todd Lipcon added a comment -

        I think at one point I opened a jira that we should create this file, then move it into place once it's closed in order to avoid this issue... can't remember the ticket number

        Show
        Todd Lipcon added a comment - I think at one point I opened a jira that we should create this file, then move it into place once it's closed in order to avoid this issue... can't remember the ticket number
        Hide
        stack added a comment -

        I couldn't find it after a depressing scroll through all the issues where you are the reporter Todd (I'm off to slash my wrists now). So, move into place, it'll either succeed or not. Ok. Sounds right. Thanks.

        Show
        stack added a comment - I couldn't find it after a depressing scroll through all the issues where you are the reporter Todd (I'm off to slash my wrists now). So, move into place, it'll either succeed or not. Ok. Sounds right. Thanks.
        Hide
        stack added a comment -

        Bringing into 0.90.x. This is kinda bad. Once we hit it, we can't online the effected region.

        Show
        stack added a comment - Bringing into 0.90.x. This is kinda bad. Once we hit it, we can't online the effected region.
        Hide
        stack added a comment -

        Does simple create in the region dir of .regioninfo. Adds timestamp to tmpfile as primitive guard against filename classes in tmpdir. After successful write, does rename.

        Show
        stack added a comment - Does simple create in the region dir of .regioninfo. Adds timestamp to tmpfile as primitive guard against filename classes in tmpdir. After successful write, does rename.
        Hide
        stack added a comment -

        I tried this up on cluster. Looks to be doing right thing. Review please.

        Show
        stack added a comment - I tried this up on cluster. Looks to be doing right thing. Review please.
        Hide
        Todd Lipcon added a comment -

        Why just WARN and not throw an exception if it can't rename? Other than that looks good.

        Show
        Todd Lipcon added a comment - Why just WARN and not throw an exception if it can't rename? Other than that looks good.
        Hide
        stack added a comment -

        I was thinking its not end of world if we don't write the .regioninfo. If we throw an exception, then region will never open.

        Show
        stack added a comment - I was thinking its not end of world if we don't write the .regioninfo. If we throw an exception, then region will never open.
        Hide
        stack added a comment -

        Committed to branch and trunk.

        Show
        stack added a comment - Committed to branch and trunk.
        Hide
        Hudson added a comment -

        Integrated in HBase-TRUNK #1737 (See https://hudson.apache.org/hudson/job/HBase-TRUNK/1737/)

        Show
        Hudson added a comment - Integrated in HBase-TRUNK #1737 (See https://hudson.apache.org/hudson/job/HBase-TRUNK/1737/ )

          People

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

            Dates

            • Created:
              Updated:
              Resolved:

              Development