fix: consider WS extMultiAddrs before publishing host address

[gabrielmer]

Description

When the user runs a Waku node with the flag --ext-multiaddr using a WS/WSS address, our current logic does not notice we specified a WS/WSS address and incorrectly adds the default wsHostAddress to the announcedAddresses

Changes

• Checking for announced WS/WSS addresses in extMultiAddrs before announcing the default wsHostAddress
• Adding test case accounting for the mentioned scenario

How to test

When WS/WSS is enabled and no WS/WSS ext-multiaddr provided, default WS/WSS address is published:

When WS/WSS is enabled and a WS/WSS ext-multiaddr is provided, the default WS/WSS is not published:

Issue

closes #1797

[github-actions]

You can find the image built from this PR at

quay.io/wakuorg/nwaku-pr:2122

Built from cf4485e

[gabrielmer]

Notice that currently the same case happens with external addresses that aren't WS/WSS: if an external address is set with the --ext-multiaddr flag and no extIp or dns are provided, both the host's bind address and the external multiaddress are announced.

We can add the same logic for that case. However I do think that it's better to require users to add extIp and extPort properly through its respective CLI flags instead of trying to infer them.

The way I see it is: extIp and extPort have their own flags which should be used. There's no flag for a separate WS/WSS address so it's either inferred from extIp/dns or passed through the ext-multiaddr flag.

Let me know what you think about this approach

[Ivansete-status]

Thanks for this!
I just added a comment mentioning how we could do the same (not a big deal though.)

However, shall we have in the PR description or somewhere an example on how to validate that the issue is sorted out with that? Maybe a snippet log will help as well.

Cheers!

[vpavlin]

LGTM, seems to correctly produce the list of listen addresses

[NagyZoltanPeter]

LGTM! Thank you!!!