nginx takes a long time to give access to public API

[netoneko]

Nginx log (possibly trouble with overlay network)

2020/03/15 20:26:10 [error] 6#6: *4885 connect() failed (113: No route to host) while connecting to upstream, client: 10.0.0.2, server: , request: "GET /vchains/2000/metrics HTTP/1.1", upstream: "http://10.0.2.33:8080/metrics", host: "52.60.147.59"
10.0.0.2 - - [15/Mar/2020:20:26:10 +0000] "GET /vchains/2000/metrics HTTP/1.1" 502 173 "-" "-" "-"
2020/03/15 20:26:10 [error] 6#6: *4893 connect() failed (113: No route to host) while connecting to upstream, client: 10.0.0.2, server: , request: "GET /vchains/2000/metrics HTTP/1.1", upstream: "http://10.0.2.33:8080/metrics", host: "52.60.147.59"

Recovers by itself, but still raises concerns. Maybe giving a shorter timeout to the resolver in the nginx config should make the problem go away.

[netoneko]

Maybe it's actually connectivity problem between the Docker swarm nodes, needs more tests:

https://stackoverflow.com/questions/58632940/docker-swarm-overlay-connect-no-route-to-host