resendDelay and AlertManager failures

[Callum Jones]

Hi there,

I was wondering if there is any documentation on how Prometheus Rulers handle the complete loss or unavailability of AlertManager nodes (basically when the notify function is unable to get a successful response from any AM node).

I see in alerting.go that there is the notion of resendDelay, my read from this is that the Ruler will keep notifying about an active alert (regardless of success) every <resendDelay> until it expires which is 4x <resendDelay> and then it would wait until the next evaluation period?

Thanks!
Callum