Our organisation uses this construct to provision ephemeral self-hosted runners.

While it works most of the time without hassle, occassionally we see jobs waiting forever for a runner that doesn't show up.

Looking through Github's webhook delivery history we observed occassional failures with various errors.

For some of them we found corresponding error from the webhook handler like I described in this issue reply. But others there doesn't seem to be matching error in the lambda function - and I suspect they are temporary issues either on the Github side or the network between Github and our AWS account.

Github's documentation describes webhook failures and recommendation on how to handle it, and I've successfully implemented it in our stack.

@kichik do you think it'll be a good idea to incorporate this into your library?