v1.0.20 introduced occasional panic when connecting to non-available Redis Cluster

[Alexander Emelin]

Hello @rueian !

Found a regression in rueidis while testing centrifuge. It triggers a panic when initializing rueidis.Client while Redis Cluster is not available. Before v1.0.20 there was no panic.

To reproduce, add test like this:

func TestConnectToNonAvailableCluster(t *testing.T) {
	_, err := NewClient(ClientOption{
		InitAddress: []string{"127.0.0.1:3000", "127.0.0.1:3001", "127.0.0.1:3002"},
	})
	if err == nil {
		t.Fatal("expected connect error")
	}
}

Do not run Redis Cluster at all (!)

Then with the help of stress tool:

go test -c
stress ./rueidis.test -test.run TestConnectToNonAvailableCluster

You should quickly see panic like this:

panic: runtime error: invalid memory address or nil pointer dereference
[signal SIGSEGV: segmentation violation code=0x2 addr=0x0 pc=0x1009ec070]

goroutine 52 [running]:
github.com/redis/rueidis.(*clusterClient).lazyRefresh(...)
	/Users/alexander.emelin/centrifugal/rueidis/cluster.go:174
github.com/redis/rueidis.newClusterClient.func1.1({0x0?, 0x0?})
	/Users/alexander.emelin/centrifugal/rueidis/cluster.go:118 +0x60
github.com/redis/rueidis.(*mux)._pipe(0x140000ce790, 0x0)
	/Users/alexander.emelin/centrifugal/rueidis/mux.go:173 +0x47c
github.com/redis/rueidis.(*mux).pipe(...)
	/Users/alexander.emelin/centrifugal/rueidis/mux.go:191
github.com/redis/rueidis.(*mux).Version(0x0?)
	/Users/alexander.emelin/centrifugal/rueidis/mux.go:205 +0x20
github.com/redis/rueidis.getClusterSlots({0x100c9a878, 0x140000ce790})
	/Users/alexander.emelin/centrifugal/rueidis/cluster.go:191 +0x4c
github.com/redis/rueidis.(*clusterClient)._refresh.func1({0x100c9a878?, 0x140000ce790?})
	/Users/alexander.emelin/centrifugal/rueidis/cluster.go:216 +0x34
created by github.com/redis/rueidis.(*clusterClient)._refresh in goroutine 34
	/Users/alexander.emelin/centrifugal/rueidis/cluster.go:215 +0x12c8

Could you please take a look?

[Rueian]

Hi @FZambia,

Wow, that's critical. Thank you so much for finding this. The fix will be merged soon.

[Alexander Emelin]

Many thanks!