Build 500s randomly on Mac rancher-desktop + dockerd

[vincent99]

ERRO[0005] Failed to build /tmp/acorn-test-app/Acornfile: Get "https://127.0.0.1:5001/v2/": http: server gave HTTP response to HTTPS client; Get "http://127.0.0.1:5001/v2/": websocket: bad handshake: error dialing backend: EOF

Usually just waiting a second and running again works. @cloudnautique says he sometimes has to delete the buildkit container to get it back.

[cjellick]

Discussed some yesterday. I think @vincent99 @cloudnautique were going to report back on whether this happens under some new conditions?

[cjellick]

We are going to refactor the buildkit pod to be exposed via a service LB rather than a proxy. Believe this will fix it

[cimtico]

I'm also running into this using k3s. It is intermittent with

ERRO[0699] Failed to run/update app: Get "https://127.0.0.1:30356/v2/": http: server gave HTTP response to HTTPS client; Get "http://127.0.0.1:30356/v2/": websocket: bad handshake: {"kind":"Status","apiVersion":"v1","metadata":{},"status":"Failure","message":"error dialing backend: EOF","code":500}

and this other error also

W1005 22:13:09.790847    4633 logging.go:59] [core] [Channel #29 SubChannel #30] grpc: addrConn.createTransport failed to connect to {
  "Addr": "/run/buildkit/buildkitd.sock",
  "ServerName": "localhost",
  "Attributes": {},
  "BalancerAttributes": null,
  "Type": 0,
  "Metadata": null
}. Err: connection error: desc = "transport: Error while dialing websocket: bad handshake: error dialing backend: EOF\n"
ERRO[0037] Failed to build /home/dmatamoros/development/hello_world/Acornfile: failed to list workers: Unavailable: connection error: desc = "transport: Error while dialing websocket: bad handshake: error dialing backend: EOF\n"

but as vincent99 mentioned, giving it a couple seconds and a couple retries later, it seems to work.

[cjellick]

Thanks for the report. We are planning to address this in the v0.4.0 release (targeted for end of this month).