10g, blocking session is null when locking occurs on other node.

I am testing a new 10g RAC (2 node) system on HP-UX.  I am testing how to 
diagnose locking issues. On node 1, I update a table. In 
another session on node 1, I update the same table and this session waits as 
expected. In v$session I can see the enq TX wait and now in 10g I can even see 
the session that is causing the block (in the blocking_session column - very 
cool). If I open a session on node 2, and also update that same table (as on 
node 1), I can see the enq TX wait, but the blocking_session status is null and 
the blocking_session_status is 'GLOBAL'. It seems to me that the 
blocking_session is missing on this second node during an TX situation. 
When this situation occurs, I can easily see that a session is being blocked 
... if it has status of GLOBAL, I can assume it is being blocked by a session 
on the other node, but I have no way of knowing what SID is causing the block. 

Is this expected behaviour?  I opened a TAR with Oracle and they could 
reproduce, but have not said if this was expected or not.

TIA.

John



   John Fedock
   "K" Line America


--
http://www.freelists.org/webpage/oracle-l

Other related posts: