nats.ConnectedHandler is not being called

[mojixcoder]

Observed behavior

After running a nats test server using this package github.com/nats-io/nats-server/v2/test and connecting to the server in tests using a code like this I saw that nats.ConnectHandler is not called after the initial connection is established. I expected to panic but the test doesn't panic

opts := []nats.Option{
	nats.PingInterval(5*time.Second),
	nats.MaxPingsOutstanding(5),
	nats.MaxReconnects(-1),
	nats.RetryOnFailedConnect(true),
	nats.ReconnectHandler(func(c *nats.Conn) {
		slog.Info("successfully reconnected to NATS")
	}),
	nats.ConnectHandler(func(c *nats.Conn) {
		panic("called")
	}),
	nats.ReconnectErrHandler(func(c *nats.Conn, err error) {
		slog.Error("error in reconnecting to NATS", slog.String("error", err.Error()))
	}),
	nats.DisconnectHandler(func(c *nats.Conn) {
		slog.Info("successfully disconnected from NATS")
	}),
}

conn, err := nats.Connect(url, opts...)

Expected behavior

Test should panic

Server and client version

Server: github.com/nats-io/nats-server/v2 v2.11.3
Client: github.com/nats-io/nats.go v1.41.2

Host environment

No response

Steps to reproduce

No response

[piotrpio]

Hello @mojixcoder, thank you for creating the issue. Is this test failing consistently or is it flaky? Could you provide a reproducible example? I tried your options in a test with embedded server:

func Test1870(t *testing.T) {
	opts := []nats.Option{
		nats.PingInterval(5 * time.Second),
		nats.MaxPingsOutstanding(5),
		nats.MaxReconnects(-1),
		nats.RetryOnFailedConnect(true),
		nats.ReconnectHandler(func(c *nats.Conn) {
			slog.Info("successfully reconnected to NATS")
		}),
		nats.ConnectHandler(func(c *nats.Conn) {
			panic("called")
		}),
		nats.ReconnectErrHandler(func(c *nats.Conn, err error) {
			slog.Error("error in reconnecting to NATS", slog.String("error", err.Error()))
		}),
		nats.DisconnectHandler(func(c *nats.Conn) {
			slog.Info("successfully disconnected from NATS")
		}),
	}

	s := RunDefaultServer()
	defer s.Shutdown()

	conn, err := nats.Connect(nats.DefaultURL, opts...)
	if err != nil {
		t.Fatalf("Unexpected error: %v", err)
	}
	defer conn.Close()
}

And it panics as expected:

go test -run Test1870 ./test/...                                                             Mon May  5 15:07:21 2025
panic: called

goroutine 12 [running]:
github.com/nats-io/nats.go/test.Test1870.func2(0x14000070720?)
        /Users/piotrpiotrowski/go/nats.go/test/conn_test.go:1318 +0x2c
github.com/nats-io/nats%2ego.Options.Connect.func1()
        /Users/piotrpiotrowski/go/nats.go/nats.go:1638 +0x28
github.com/nats-io/nats%2ego.(*asyncCallbacksHandler).asyncCBDispatcher(0x140001d0d60)
        /Users/piotrpiotrowski/go/nats.go/nats.go:3074 +0x28
created by github.com/nats-io/nats%2ego.Options.Connect in goroutine 7
        /Users/piotrpiotrowski/go/nats.go/nats.go:1635 +0x35c
FAIL    github.com/nats-io/nats.go/test 0.282s
FAIL

So more details would be appreciated. Thanks!

[mojixcoder]

Here it is:

import natstestserver "github.com/nats-io/nats-server/v2/test"

const (
	natsUsername = "username"
	natsPassword = "pass1234"
)

func newTestNATSServer(t *testing.T) string {
	opts := natstestserver.DefaultTestOptions
	opts.Port = -1
	opts.Username = natsPassword
	opts.Password = natsPassword

	server := natstestserver.RunServer(&opts)

	t.Cleanup(server.Shutdown)

	return server.ClientURL()
}

func Test1870(t *testing.T) {
	address := newTestNATSServer(t)

	opts := []nats.Option{
		nats.PingInterval(5 * time.Second),
		nats.MaxPingsOutstanding(5),
		nats.MaxReconnects(-1),
		nats.RetryOnFailedConnect(true),
		nats.UserInfo(natsUsername, natsPassword),
		nats.ReconnectHandler(func(c *nats.Conn) {
			slog.Info("successfully reconnected to NATS")
		}),
		nats.ConnectHandler(func(c *nats.Conn) {
			panic("connected")
		}),
		nats.ReconnectErrHandler(func(c *nats.Conn, err error) {
			slog.Error("error in reconnecting to NATS", slog.String("error", err.Error()))
		}),
		nats.DisconnectHandler(func(c *nats.Conn) {
			slog.Info("successfully disconnected from NATS")
		}),
	}

	_, err := nats.Connect(address, opts...)
	assert.NoError(t, err)
}

And it consistently never fails.

I went into nats.Connect() source code and I saw that connectionEstablished boolean is false.

EDIT:
Ok after removing these configs: nats.MaxReconnects(-1), nats.RetryOnFailedConnect(true) I'm getting this error: nats: Authorization Violation and I don't know why.

Another question: if MaxReconnects = -1 and RetryOnFailedConnect = true, if the connection doesn't get established at the first try, it doesn't return any errors nor calls ConnectHandler. How about if it got connected in nth reconnection and it is the first time that a connection gets established, does it call ConnectHandler in this case?

[piotrpio]

Ok, so I think I figured it out. In newTestNATSServer, you're setting invalid username:

	opts.Username = natsPassword
	opts.Password = natsPassword

should be:

	opts.Username = natsUsername
	opts.Password = natsPassword

If you use RetryOnFailedConnect(true), the test passes because client ignores the failed attempt and just tries again (without returning error from Connect(). Test finishes right after the first attempt so it passes just fine, even though the connection is not established. After removing RetryOnFailedConnect() from opts, the test fails with Authorization Violation (as expected).

However, this uncovers a different issue. I would assume, that this initial connect error should cause ReconnectErrHandler to be invoked (with authorization violation error). But it's not, turns out ReconnectErrHandler is only invoked on the subsequent failed connection attempts, which hides the first error. This I think should be fixed, I'll take care of this.