REFUSED CONNECTION TO BorsNG.UserSocket after upgrading

Support
#1

I've got a private bors instance on heroku, upgraded from 264fe462..HEAD (master today). I'm now getting errors when the github webhook tries to send data locally:

2020-03-06T18:42:53.893137+00:00 app[web.1]: 18:42:53.892 pid=<0.531.0> [info] REFUSED CONNECTION TO BorsNG.UserSocket in 443µs
2020-03-06T18:42:53.893151+00:00 app[web.1]:   Transport: :websocket
2020-03-06T18:42:53.893152+00:00 app[web.1]:   Serializer: Phoenix.Socket.V2.JSONSerializer
2020-03-06T18:42:53.893152+00:00 app[web.1]:   Connect Info: %{}

I originally had a database error, where the database wasn't migrated; I manually did a ecto.migrate which resolved that; but now I'm also seeing intermittent:

2020-03-06T18:43:00.278823+00:00 app[web.1]: 18:43:00.278 request_id=41fe6054-db93-4c0b-95f0-2804b6adb309 pid=<0.536.0> [info] POST /webhook/github
2020-03-06T18:43:00.282555+00:00 app[web.1]: 18:43:00.281 request_id=41fe6054-db93-4c0b-95f0-2804b6adb309 pid=<0.536.0> [info] Sent 500 in 2ms
2020-03-06T18:43:00.282729+00:00 app[web.1]: 18:43:00.281 request_id=41fe6054-db93-4c0b-95f0-2804b6adb309 pid=<0.536.0> [info] Converted error {:case_clause, "neutral"} to 500 response
2020-03-06T18:43:00.283665+00:00 app[web.1]: 18:43:00.282 pid=<0.418.0> [error] #PID<0.536.0> running BorsNG.Endpoint (cowboy_protocol) terminated
2020-03-06T18:43:00.283666+00:00 app[web.1]: Server: ubors.herokuapp.com:80 (http)
2020-03-06T18:43:00.283667+00:00 app[web.1]: Request: POST /webhook/github
2020-03-06T18:43:00.283667+00:00 app[web.1]: ** (exit) an exception was raised:
2020-03-06T18:43:00.283668+00:00 app[web.1]:     ** (CaseClauseError) no case clause matching: "neutral"
2020-03-06T18:43:00.283669+00:00 app[web.1]:         (bors) lib/github/github.ex:297: BorsNG.GitHub.map_state_to_status/1
2020-03-06T18:43:00.283669+00:00 app[web.1]:         (bors) lib/web/controllers/webhook_controller.ex:233: BorsNG.WebhookController.do_webhook/3
2020-03-06T18:43:00.283670+00:00 app[web.1]:         (bors) lib/web/controllers/webhook_controller.ex:83: BorsNG.WebhookController.webhook/2
2020-03-06T18:43:00.283670+00:00 app[web.1]:         (bors) lib/web/controllers/webhook_controller.ex:1: BorsNG.WebhookController.action/2
2020-03-06T18:43:00.283671+00:00 app[web.1]:         (bors) lib/web/controllers/webhook_controller.ex:1: BorsNG.WebhookController.phoenix_controller_pipeline/2
2020-03-06T18:43:00.283671+00:00 app[web.1]:         (phoenix) lib/phoenix/router.ex:288: Phoenix.Router.__call__/2
2020-03-06T18:43:00.283672+00:00 app[web.1]:         (bors) lib/web/bors_ng/endpoint.ex:1: BorsNG.Endpoint.plug_builder_call/2
2020-03-06T18:43:00.283672+00:00 app[web.1]:         (bors) lib/web/bors_ng/endpoint.ex:1: BorsNG.Endpoint.call/2

What can I do to debug/fix?

#2

Prior to the migration my bors URL was just returning a "internal server error"; now it seems up, but I'm still getting the refused connection issues in the logs.

#3

I also see these REFUSED CONNECTION TO BorsNG.UserSocket messages every ~5s:

Dec 04 11:13:23 bors 8yq42s7w8537528rcbgpqp5pbgad58x3-unit-script-bors-ng-start[15958]: 11:13:23.005 pid=<0.2421.0> [info] REFUSED CONNECTION TO BorsNG.UserSocket in 197µs
Dec 04 11:13:23 bors 8yq42s7w8537528rcbgpqp5pbgad58x3-unit-script-bors-ng-start[15958]:   Transport: :websocket
Dec 04 11:13:23 bors 8yq42s7w8537528rcbgpqp5pbgad58x3-unit-script-bors-ng-start[15958]:   Serializer: Phoenix.Socket.V2.JSONSerializer
Dec 04 11:13:23 bors 8yq42s7w8537528rcbgpqp5pbgad58x3-unit-script-bors-ng-start[15958]:   Parameters: %{"token" => "XXX", "vsn" => "2.0.0"}

Should I be worried about this? (Bors seems to work correctly).

#4

If BorsNG.UserSocket isn't working, then that means you won't get the "This page is out of date [reload]" pop-up on the dashboard. It makes the dashboard marginally less useful as a way to monitor a repo, but it won't break any core functionality.