In some architectures, where many microservices are behind load balancers and other front end proxies, sometimes it is the NOC’s fault. Yes a 500 should really only be a backend error.
Oh even better.
I manage Kong (an API manager) and 90% of my time these days is explaining illiterate devs that no, Kong is fine, it's their shitty API crashing. Or better, their shitty HAproxy they don't know how to configure.
FML.
8 Comments
eager_eagle@lemmy.world · 58 pts · 1y
hey, at least it's not a 200 with
{ "error": "Bad request" }prowe45@lemm.ee · 12 pts · 1y
Yeah, plus a 500 is the backend admitting it fucked up and not blaming the front end like a 4xx would be.
slazer2au@lemmy.world · 19 pts · 1y
Quick, make a ticket to the network team and ask why the firewall is causing the internal server error.
I wish I could say I haven't seen devs do this.
jaybone@lemmy.world · 7 pts · 1y
In some architectures, where many microservices are behind load balancers and other front end proxies, sometimes it is the NOC’s fault. Yes a 500 should really only be a backend error.
fibojoly@sh.itjust.works · 2 pts · 1y
Oh even better. I manage Kong (an API manager) and 90% of my time these days is explaining illiterate devs that no, Kong is fine, it's their shitty API crashing. Or better, their shitty HAproxy they don't know how to configure. FML.
prole@lemmy.blahaj.zone · 6 pts · 1y
Fuck yes, Perry Bible Fellowship.
Best web comic of all time, hands down.
CodeMonkey@programming.dev · 3 pts · 1y
Could be worse. You could just have your socket disconnect because the back end process crashed.
Kojichan@lemmy.world · 3 pts · 1y
Except when it's a front-end script causing issues and blaming it kn the backend.