[
http://jira.jboss.com/jira/browse/JBREM-657?page=all ]
Ron Sigal closed JBREM-657.
---------------------------
Resolution: Done
The nondeterminism that resulted from putting the network i/o on a separate thread causes
too many problems, so it has been backed out.
Instead, before any network i/o is performed, the value of disconnectTImeout, set by
Client.setDisconnectTImeout(), is checked.
If (disconnectTimeout < 0) then it is ignored
else if (disconnectTimeout == 0) then no network i/o is carried out
else disconnectTimeout is use as a per invocation timeout for the network i/o.
So, if Client.disconnectTimeout == 0, disconnect() and removeListener() behave like the
no longer existent disconnectLocal() and removeListenerLocal().
Implement versions of Client.removeListener() and Client.disconnect()
which do not write to a broken server.
------------------------------------------------------------------------------------------------------------
Key: JBREM-657
URL:
http://jira.jboss.com/jira/browse/JBREM-657
Project: JBoss Remoting
Issue Type: Feature Request
Security Level: Public(Everyone can see)
Affects Versions: 2.2.0.Beta1 (Bluto)
Reporter: Ron Sigal
Assigned To: Ron Sigal
Fix For: 2.2.0.Beta1 (Bluto)
From Clebert Suconic:
When a server is killed and failover is processed, we don't have at this
point a way to disconnect a client from a dead server. I tried to call
disconnect on those clients after failover and remoting does a socket
communication.
It looks we would need a new feature/new method such as
client.removeLeases or client.disconnect(false); false means = send data
or any other similar way.
--
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators:
http://jira.jboss.com/jira/secure/Administrators.jspa
-
For more information on JIRA, see:
http://www.atlassian.com/software/jira