Nathan Sharp wrote:

> As talked about earlier in this mailing list, I've created a patch to 
> the 1.4.2 server which makes the server send blank packets to all 
> connected clients once per time period if no other client activity has 
> happened.  This allows the server to notice bum connections much 
> faster. I've tested this patch with a number of clients, but certainly 
> not them all.  Exodus 0.6.0.0 will stop working if you use this 
> feature, but there is a fix in CVS for this (so look out for the next 
> release). 
> I'm still not 100% satisfied w/ this solution.  In my testing it can 
> still take 10-20 minutes for the server to notice laptop users who 
> pull the network cable while connected.  The problem is that the 
> network tries to deliver the packets for a long time before giving 
> up.  I'd like to eventually add something to handle this, but that 
> will be more work, and probably a JEP because it'd require protocol 
> changes.  If anyone has ideas or thoughts, I'd be happy to discuss this.
>
>  Nathan

Are you thinking of having an explicit heartbeat required for clients to 
send over the network socket? I'm not sure if I agree with circumventing 
the TCP timeout with application logic or not - it seems like just 
giving a delivery error once the socket closes would be a better 
long-term solution.

-David Waite

_______________________________________________
jdev mailing list
[EMAIL PROTECTED]
http://mailman.jabber.org/listinfo/jdev

Reply via email to