[jira] Commented: (DIRSERVER-1377) Potential concurrency issue when adding/modifying/deleting entries at a high rate

    [ https://issues.apache.org/jira/browse/DIRSERVER-1377?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12722995#action_12722995 ]

Emmanuel Lecharny commented on DIRSERVER-1377:
----------------------------------------------

Update :

After having put some more logs, and grepped/sedded gigabytes of logs (no fun at all), I have found some dubious code and result. Here is the log I obtain just before the first NPE :

---> pool-6-thread-7 - ---> Remove apacheOneLevel_forward = 1, 226034
     pool-6-thread-7 - <--- Remove AVL apacheOneLevel_forward = 1, 226034    
     pool-6-thread-7 - ---> Remove apacheOneLevel_reverse = 226034
     pool-6-thread-7 - <--- Remove AVL apacheOneLevel_reverse = 226034
     pool-6-thread-7 - ---> Remove apacheSubLevel_forward = 226034, 226034
     pool-6-thread-7 - <--- Remove AVL apacheSubLevel_forward = 226034, 226034
     pool-6-thread-7 - ---> Remove apacheSubLevel_reverse = 226034
     pool-6-thread-7 - <--- Remove AVL apacheSubLevel_reverse = 226034
---> pool-6-thread-7 - ---> Remove apacheOneLevel_forward = 1, 226034
     pool-6-thread-7 - Error while removing 1, 226034 on table apacheOneLevel_forward
     java.lang.NullPointerException
        at org.apache.directory.server.core.avltree.AvlTreeMarshaller.deserialize(AvlTreeMarshaller.java:240)

As one can see, we remove twice something from the oneLevelIndex. The code which does that is :

        ndnIdx.drop( id );
        updnIdx.drop( id );
*       oneLevelIdx.drop( id );
        entryCsnIdx.drop( id );
        entryUuidIdx.drop( id );

        if( id != 1 )
        {
            subLevelIdx.drop( id );
        }
       
        // Remove parent's reference to entry only if entry is not the upSuffix
        if ( !parentId.equals( 0L ) )
        {
*           oneLevelIdx.drop( parentId, id );
        }

Sadly (?) the second removal has no impact (I checked it), except sucking CPU.

Otherwise, *all* the exceptions I get are from the same line of code : AvlTreeMarshaller.java:240

             for( int i = 0; i < nodes.length - 1; i++ )
            {
                nodes[ i ].setNext( nodes[ i + 1] );
*              nodes[ i + 1].setPrevious( nodes[ i ] );
            }

which means the index table is broken (nodes[i+1] is null). There are two possibilities for this table to be incorrect :
- we save some bad table
- we recreate a bad table

I will double check that code.

Otherwise, we discussed a lot with Kiran about serialization/deserialization, and we agreed that it should be done *inside* jdbm. Will work on that too.

--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.