We have one server, and one client attached to the channel. The channel, 
server, and client are all running on the same machine. Occasionally after the 
client is killed the channel crashes with the following stack trace:

ACE_Guard<ACE_Lock>::acquire() line 9 + 10 bytes
ACE_Guard<ACE_Lock>::ACE_Guard<ACE_Lock>(ACE_Lock & {...}) line 36
TAO_Connection_Handler::decr_pending_upcalls() line 49 + 15 bytes
TAO_IIOP_Connection_Handler::handle_close(void * 0xffffffff, unsigned long 511) 
line 206 + 11 bytes
TAO_Transport::close_connection_i() line 682
TAO_Transport::close_connection() line 639
TAO_GIOP_Invocation::close_connection() line 451
TAO_ORB_Core::service_raise_comm_failure(TAO_GIOP_Invocation * 0x0156f830, 
TAO_Profile * 0x00eafe60, CORBA_Environment & {...}) line 1399
TAO_GIOP_Synch_Invocation::invoke_i(unsigned char 0, CORBA_Environment & {...}) 
line 677 + 41 bytes
TAO_GIOP_Twoway_Invocation::invoke(TAO_Exception_Data * 0x101054b0 
_tao_CosEventComm_PushConsumer_push_exceptiondata, unsigned int 1, 
CORBA_Environment & {...}) line 834 + 17 bytes
CosEventComm::_TAO_PushConsumer_Remote_Proxy_Impl::push(CORBA_Object * 
0x00eafce0, const CORBA_Any & {...}, CORBA_Environment & {...}) line 702 + 23 
bytes
CosEventComm::PushConsumer::push(const CORBA_Any & {...}, CORBA_Environment & 
{...}) line 1141
TAO_CEC_ProxyPushSupplier::push_to_consumer(const CORBA_Any & {...}, 
CORBA_Environment & {...}) line 321
TAO_CEC_Push_Command::execute(CORBA_Environment & {...}) line 102
TAO_CEC_Dispatching_Task::svc() line 36 + 15 bytes
ACE_Task_Base::svc_run(void * 0x00e829b0) line 203 + 11 bytes
ACE_Thread_Adapter::invoke_i() line 148 + 7 bytes
ACE_Thread_Adapter::invoke() line 91 + 11 bytes
ace_thread_adapter(void * 0x00e8d808) line 121 + 10 bytes
_threadstartex(void * 0x00e8d870) line 212 + 13 bytes
KERNEL32! 77e887dd()

The server stays connected when the client goes up and down.

When the event service is run on a seperate computer, the problem does not seem 
to exist.