[zeromq-dev] TIME_WAIT state when using source address in a TCP connection

[Jim Melton]

This seems related to https://github.com/zeromq/pyzmq/issues/1088
--
Jim Melton




> On Mar 29, 2022, at 7:48 AM, Tom Bequin <tom.bequin at studio-nyx.com> wrote:
> 
> Hello,
> 
> We are establishing connections in TCP without any problem.
> For that, we have a server binding to a local address of model "tcp://192.168.1.1:49000 <http://192.168.1.1:49000/>".
> We connect a client to a distant address of model "tcp://192.168.1.1:49000 <http://192.168.1.1:49000/>".
> 
> When using what is called in the source code "source address", to connect using a different interface, as of model "tcp://192.168.1.2:49000 <http://192.168.1.2:49000/>;192.168.1.1:49000 <http://192.168.1.1:49000/>", the connection is working well, but the connected socket remains in a TIME_WAIT state so the next connection fails with port reused.
> 
> This is not the case with standard TCP messages, so why is it happening only in a case of a source address?
> 
> The SO_LINGER option is well set to {1, 0}. We are using Windows 10.
> 
> May you be of any help?
> 
> Tom Bequin,
> Dev at NYX STUDIO.
> _______________________________________________
> zeromq-dev mailing list
> zeromq-dev at lists.zeromq.org
> https://lists.zeromq.org/mailman/listinfo/zeromq-dev

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.zeromq.org/pipermail/zeromq-dev/attachments/20220402/7c0cc223/attachment.htm>