protocols/dcutr/examples: Set libp2p-ping to keep conn alive

[mxinden]

Description

The example is for testing out the hole punching capabilities only. Thus best to
remove the complexity of connection keep alives via the ping protocol.

Links to any relevant issues

Might make #2609 (comment) easier to debug. //CC @T2JOESl4m2ZpNC

Change checklist

• I have performed a self-review of my own code
• I have made corresponding changes to the documentation
• I have added tests that prove my fix is effective or that my feature works
• A changelog entry has been made in the appropriate crates

[thomaseizinger]

I am okay with the change but confused by your PR description.

Thus best to remove the complexity of connection keep alives via the ping protocol.

You are talking about removing but the diff says you are activating keep_alive?

[thomaseizinger]

Ah, I think it just clicked for me.

I think what is meant is: Keeping a connection alive without ping is complex and by activating ping's keep-alive feature, we remove this complexity.

[elenaf9]

Not sure if I understand this change. How is not keeping the connection alive currently causing additional complexity?
As a listener, we keep a long-lived connection to the relay anyway, and as a dialer we only want to try hole-punching to the target. After the hole-punch we exchange Identify events and I think it makes sense that the connection eventually closes again.
With the keep alive as a dialer we would keep the connection to the relay, and even after both client successfully hole-punched they would still also keep their relayed connection alive. If someone would be using this example as a reference it might cause some unwanted behaviour for them.

[mxinden]

With the keep alive as a dialer we would keep the connection to the relay, and even after both client successfully hole-punched they would still also keep their relayed connection alive. If someone would be using this example as a reference it might cause some unwanted behaviour for them.

True, though users seem to be confused that they get a DirectConnectionUpgradeSucceeded followed by a ConnectionClosed.

I don't have a strong opinion here. You have a good point above. As a compromise, how about calling out that one would not use keep_alive(true) in production code?