The method used for dialing addresses configured as persistent peers is the following:
|
func (sw *Switch) reconnectToPeer(addr *NetAddress) { |
|
if sw.reconnecting.Has(string(addr.ID)) { |
|
return |
|
} |
|
sw.reconnecting.Set(string(addr.ID), addr) |
|
defer sw.reconnecting.Delete(string(addr.ID)) |
|
|
|
start := time.Now() |
|
sw.Logger.Info("Reconnecting to peer", "addr", addr) |
|
for i := 0; i < reconnectAttempts; i++ { |
|
if !sw.IsRunning() { |
|
return |
|
} |
|
|
|
err := sw.DialPeerWithAddress(addr) |
|
if err == nil { |
|
return // success |
|
} else if _, ok := err.(ErrCurrentlyDialingOrExistingAddress); ok { |
|
return |
|
} |
|
|
|
sw.Logger.Info("Error reconnecting to peer. Trying again", "tries", i, "err", err, "addr", addr) |
|
// sleep a set amount |
|
sw.randomSleep(reconnectInterval) |
|
continue |
|
} |
|
|
|
sw.Logger.Error("Failed to reconnect to peer. Beginning exponential backoff", |
|
"addr", addr, "elapsed", time.Since(start)) |
|
for i := 0; i < reconnectBackOffAttempts; i++ { |
|
if !sw.IsRunning() { |
|
return |
|
} |
|
|
|
// sleep an exponentially increasing amount |
|
sleepIntervalSeconds := math.Pow(reconnectBackOffBaseSeconds, float64(i)) |
|
sw.randomSleep(time.Duration(sleepIntervalSeconds) * time.Second) |
|
|
|
err := sw.DialPeerWithAddress(addr) |
|
if err == nil { |
|
return // success |
|
} else if _, ok := err.(ErrCurrentlyDialingOrExistingAddress); ok { |
|
return |
|
} |
|
sw.Logger.Info("Error reconnecting to peer. Trying again", "tries", i, "err", err, "addr", addr) |
|
} |
|
sw.Logger.Error("Failed to reconnect to peer. Giving up", "addr", addr, "elapsed", time.Since(start)) |
|
} |
According with the comments in the code, and some of our documentation, the total duration of re-connection attempts should be around 1 day:
|
// repeatedly try to reconnect for a few minutes |
|
// ie. 5 * 20 = 100s. |
|
reconnectAttempts = 20 |
|
reconnectInterval = 5 * time.Second |
|
|
|
// then move into exponential backoff mode for ~1day |
|
// ie. 3**10 = 16hrs. |
|
reconnectBackOffAttempts = 10 |
|
reconnectBackOffBaseSeconds = 3 |
But, from this snap of the code, derived from the previous one, the total interval between the configured attempts (30) is actually around 8h14m: https://go.dev/play/p/JWfU6lerps5.
Namely, the expected behavior does not match the implemented behavior.