On 02/09/2016 11:52 AM, Jacob Hoffman-Andrews wrote:
> As I said previously, I think it would be better for implementers to
> query each IP they receive, until they get a success.
I thought some more about this from a CA implementation perspective, and
it would actually be rather painful and error-prone to implement.

In Boulder, we have a maximum amount of time we are willing to spend on
validating a challenge. Currently it's 60 seconds. We may increase it at
some point, but there will always be some limit. For hostnames that
return a large number of IP addresses, it's entirely possible we would
timeout before reaching the one IP address that is provisioned with the
challenge.

That means that instead of the nice clean "Push a challenge to any
server and it will work" guarantee, we would have a "Push a challenge to
any server, and it will work provided you don't have too many IP
addresses and the other instances respond quickly enough" guarantee.

_______________________________________________
Acme mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/acme

Reply via email to