We deployed an internal app that routes through our DNS and VPN, so all of our traffic went from all over the US to Phoenix, AZ. From there, it looks like LAX for the edge node, then to Denver, CO, for hosting.
The devices that were part of this deployment were kiosks when we deployed the new app, several of the kiosks got 502 errors, then after a refresh, would land on the new site. Then every 5s, the site refreshes so the data is up-to-date. Some clients are old, so we could not use js to fetch new data. Eventually, even the kiosks that migrated successfully eventually got the 502. We can only track the 502 at the edge. The error never made it to our instances. Any idea what could be causing the issues? The servers handled the traffic beautifully, it just seems like some of the traffic never made it there.