# TCP connection pool: better separation

**URL:** <https://discourse.nodered.org/t/tcp-connection-pool-better-separation/19432>\
**Category:** Feature Requests\
**Created:** [21 December 2019 17:54 UTC](https://discourse.nodered.org/t/tcp-connection-pool-better-separation/19432 "2019-12-21T17:54:55Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![dvv](https://sea2.discourse-cdn.com/flex026/user_avatar/discourse.nodered.org/dvv/32/15374_2.png) [@dvv](https://discourse.nodered.org/u/dvv)\
**Post date:** [21 December 2019 17:54 UTC](https://discourse.nodered.org/t/tcp-connection-pool-better-separation/19432/1 "2019-12-21T17:54:55Z")

</div>

Hello list!

Today I learnt that if one feed TCP reply node with message without `_session` it will go to _all_ client connections across _the whole_ project, not limited with current flow.

This fact is indeed written in the node help and I knew that, but so far I had only one TCP gate in my project and it's been behaving well.

By TCP gate I mean a TCP In node which maps incoming messages to my flow and maps messages from the flow to TCP Out Reply node. When a message comes from the flow it has no `_session` and naturally goes to all connected clients.

However, now I have two such TCP gates and I must distinguish all connected clients to gate A from all connected clients to gate B.

So basically I'd like to extend `_session` to contain another identifier -- I believe TCP In node name could go well. So that messages unaware of particular `_session.id` could still be routed to subset of `connectionPool` by providing `_session.name`. This change should not impact existings flows.

If it is unwanted to touch anything in the core, please explain how to _reliably_ keep actual list of connected clients of a specific TCP In node in context? Given `connectionPool` is hidden and status node does not provide session info upon `common.status.[dis]connected`.

TIA,  
--Vladimir