Closed laforge49 closed 11 years ago
"On the destination node, it was also found that the to actor pends when doing a jid copy until the halt actor returns control, despite the halt actor having its own async mailbox."
ConsoleAgent.out was sharing the same mailbox as the agent itself, which is why when making a copy of out, processing of the copy request was waiting for the agent to return control.
ConsoleAgent.out now has its own mailbox.
"Combining the to and halt commands results in the sending node locking up."
The saved RP objects were being accessed outside of the context of the AgentChannel actor, by way of a close method. Replaced the method call to close with sending a CloseChannel event.
Combining the to and halt commands results in the sending node locking up.
On the destination node, it was also found that the to actor pends when doing a jid copy until the halt actor returns control, despite the halt actor having its own async mailbox.