Subsequent to seeing a very small number of validation timeouts over the weekend... I had another think about this.
We have the Scoop API configured such that its gunicorn will send us a 502 in 60s, if the app hasn't replied at that point. We should therefore wait the full 60s plus a small buffer here... because if we don't, then we aren't distinguishing between "Scoop API hit its limit" and "Scoop API is unresponsive".
I'd like to see them distinguished, so that I get a better sense of whether we need to try and enforce a timeout on gethostbyname.
Subsequent to seeing a very small number of validation timeouts over the weekend... I had another think about this.
We have the Scoop API configured such that its gunicorn will send us a 502 in 60s, if the app hasn't replied at that point. We should therefore wait the full 60s plus a small buffer here... because if we don't, then we aren't distinguishing between "Scoop API hit its limit" and "Scoop API is unresponsive".
I'd like to see them distinguished, so that I get a better sense of whether we need to try and enforce a timeout on
gethostbyname.