Help troubleshooting rogue machine connection

My app never puts all of the machines in to suspend mode. Looking at the live logs, it appears a request is made to my service every minute, so obviously this is the culprit. However, I don’t know how I can find where this request is coming from. My service is a Remix application. I previously had Health Checks enabled, but disabled those while trying to debug this problem. Everything used to work fine, but I can see on 1/19 is when the consistent requests started happening. I had a small UI change and deployment on this date, but nothing that should have impacted this. This is impacting my beta instance only, and not my prod instance which has the same code and config deployed after the 19th. Any help trying to troubleshoot where the connection is coming from or killing that connection would be much appreciated. Thanks!

My current config

deploy = { }
primary_region = "sea"

[http_service]
auto_start_machines = true
auto_stop_machines = "suspend"
force_https = true
internal_port = 3_000
min_machines_running = 0
processes = [ "app" ]

[[vm]]
cpu_kind = "shared"
cpus = 1
memory_mb = 1_024

My old config

primary_region = "sea"

[build]

[deploy]
  strategy = "canary"

[http_service]
  internal_port = 3000
  force_https = true
  auto_stop_machines = "suspend"
  auto_start_machines = true
  min_machines_running = 0
  processes = ["app"]

  [[http_service.checks]]
    grace_period = "10s"
    interval = "15m"
    method = "GET"
    timeout = "10s"
    path = "/api/fly/healthcheck"

[[vm]]
  cpu_kind = "shared"
  cpus = 1
  memory_mb = 1024

I know nothing of Remix but I’m sure it - or the HTTP server that is serving it - has an access log of some sort where you could check the HTTP requests coming in…?

that would make it easy to see details about the request coming in every minute (depending on the logging level, I guess), before wasting hours in spelunking your app/infra setup only to discover there is an external uptime monitor service pinging you every 60 seconds :sweat_smile:

Good call. I’ll see what I can uncover there. I only have Sentry hooked up for now

Wow, so it looks like uptime Monitoring was configured in Sentry and only for the beta app, and they do this automatically. Thanks for the help!

:smile: I was writing from experience (including and especially the “wasting hours” part), glad it was helpful