should there be a parameter for which mode is expected? Could you have both an event and poll mode with a trigger body (event using it just to trigger once?)

@ssoroka currently handshake_body is sent after establishing connection (once). The trigger_body is sent on each gather cycle, so currently I use trigger_body as switch between using Gather() (sampling) and Start() with listen() (event-based) for acquisition. The reason is that I have two use-cases, one where we need to send a handshake-message and then receive data upon change of data and another one, where we need to send a handshake-message (like registering the client) and then a trigger-message to trigger sending of data.
We even might cover a third use-case if we allow multiple handshake-messages...

What do you suggest?

!retry-failed

Hi Sven, I just got a chance to look at this. I like it and have some questions. How do you intend to use this input? Are you going to write a custom websocket server or use an existing service? Many websocket servers require authentication and have a custom request protocol that might be too complex for the trigger_body mechanism.

I also think it would be good to switch this to gorilla/websocket so telegraf uses only one websocket library. I don't see any other plugins that use gorilla/websocket currently but some of our module dependencies use it, and the recent websocket output uses it.

@reimda I use it for an existing service that can be handled with the means provided. Another important use-case is to connect two telegrafs in restricted (NATed) networks. Using the new websocket output plugin, we can connect two instances where an "inner" telegraf has to establish connects to an "outer" telegraf. Kind of a proxy setups with connection restrictions in place...
However, I agree that this cannot cover arbitrary protocols, but it will be sufficient for the two use-cases above.

Currently, converting this to gorilla is of low priority here due to limited time. So if anybody (looks at @FZambia ;-)) wants to step up on it, any help is appreciated. :-)

I use it for an existing service that can be handled with the means provided. Another important use-case is to connect two telegrafs in restricted (NATed) networks. Using the new websocket output plugin, we can connect two instances where an "inner" telegraf has to establish connects to an "outer" telegraf.

It would be useful to add some usage examples to the readme. Maybe these would be good examples to add if you get a chance. Thanks!

Will close this as there seems to be no great interest in this PR. If anyone wants to continue here, please feel free to take the code and work on it.