net.spy.MemcachedClient shutdown exceptions

[David Olliges]

Hi,

I'm seeing intermittent Exceptions in the logs when running a test  
that creates a new client, uses it, then shuts it down.

Exception in thread "Memcached IO over {MemcachedConnection to  
localhost/127.0.0.1:11221 localhost/127.0.0.1:11222} - SHUTTING DOWN  
(informed client)" java.nio.channels.CancelledKeyException
	at sun.nio.ch.SelectionKeyImpl.ensureValid(SelectionKeyImpl.java:55)
	at sun.nio.ch.SelectionKeyImpl.interestOps(SelectionKeyImpl.java:64)
	at net.spy.memcached.protocol.TCPMemcachedNodeImpl.fixupOps 
(TCPMemcachedNodeImpl.java:372)
	at net.spy.memcached.MemcachedConnection.handleIO 
(MemcachedConnection.java:275)
	at net.spy.memcached.MemcachedConnection.handleIO 
(MemcachedConnection.java:182)
	at net.spy.memcached.MemcachedClient.run(MemcachedClient.java:740)

Exception in thread "Memcached IO over {MemcachedConnection to  
localhost/127.0.0.1:11221 localhost/127.0.0.1:11222} - SHUTTING DOWN  
(telling client)" java.lang.NullPointerException
	at net.spy.memcached.protocol.TCPMemcachedNodeImpl.fixupOps 
(TCPMemcachedNodeImpl.java:372)
	at net.spy.memcached.MemcachedConnection.handleIO 
(MemcachedConnection.java:275)
	at net.spy.memcached.MemcachedConnection.handleIO 
(MemcachedConnection.java:182)
	at net.spy.memcached.MemcachedClient.run(MemcachedClient.java:740)

Exception in thread "Memcached IO over {MemcachedConnection to  
localhost/127.0.0.1:11221 localhost/127.0.0.1:11222} - SHUTTING DOWN  
(telling client)" java.lang.NullPointerException
	at net.spy.memcached.protocol.TCPMemcachedNodeImpl.fixupOps 
(TCPMemcachedNodeImpl.java:372)
	at net.spy.memcached.MemcachedConnection.handleIO 
(MemcachedConnection.java:275)
	at net.spy.memcached.MemcachedConnection.handleIO 
(MemcachedConnection.java:182)
	at net.spy.memcached.MemcachedClient.run(MemcachedClient.java:740)


When I look at the source, I don't see any code that tries to prevent  
the race condition between the IO handling code in  
MemcachedConnection and the shutdown() call that closes each  
connection.  Is this a bug, or am I misreading the code?

Thanks,
David