Extend --trace or add a --debug flag to see server connections

[philpennock]

Proposed change

At present, if there is a cluster of hosts and the natscli connects but the TLS cert fails to verify, it appears that it will use the pre-TLS INFO line to look at alternative host IPs reported and try them in turn, until one connects.

But that happens without diagnostic feedback. Which is great for self-healing resiliency, but not for cluster administrators trying to carefully watch the rollout of a change.

Use case

If doing upgrades and trying to see TLS change impact, it's good to know which server is actually being talked to. Having --debug report connection attempts would make it clearer when silent reconnections are happening. Having a --no-reconnect might help too.

Contribution

Sure

[ripienaar]

Its more that nats.go is doing those things opaquely. So dont think I can do much on the CLI side.

Best I can probably do is add the missing handlers like this - but this wont tell you what you want, we'll have to expand nats.go

+               nats.ConnectHandler(func(conn *nats.Conn) {
+                       if opts.Trace {
+                               log.Printf(">>> Connected to %s", conn.ConnectedUrlRedacted())
+                       }
+               }),
+               nats.DiscoveredServersHandler(func(conn *nats.Conn) {
+                       if opts.Trace {
+                               log.Printf(">>> Discovered new servers, known servers are now %s", strings.Join(conn.Servers(), ", "))
+                       }
+               }),

[philpennock]

I have a chatbot which talks to NATS and it has

	opts = append(opts,
		nats.ConnectHandler(func(nc *nats.Conn) {
			s.
				WithField("cluster", nc.ConnectedClusterName()).
				WithField("server-id", nc.ConnectedServerId()).
				WithField("server-name", nc.ConnectedServerName()).
				WithField("addr", nc.ConnectedAddr()).
				Info("connected")
		}),
		nats.ReconnectHandler(func(nc *nats.Conn) {
			s.
				WithField("cluster", nc.ConnectedClusterName()).
				WithField("server-id", nc.ConnectedServerId()).
				WithField("server-name", nc.ConnectedServerName()).
				WithField("addr", nc.ConnectedAddr()).
				Info("reconnected")
		}),
		nats.DisconnectErrHandler(func(nc *nats.Conn, err error) {
			s.WithError(err).Warn("disconnected")
		}),
/* ... */

That's comprehensive enough for my diagnostics logging and about what I'm after here, I think.