71754a48-09bb-4250-b361-ff5ab1df9597

Found a message that might explain the problem (edit: not sure)
https://discord.com/channels/713503345364697088/1210336429855473705/1210697666514915368
So since it connects to an external/public database it can't go to sleep, even without outbound traffic?

Also it seems it's only incoming traffic which should not matter for App Sleeping according to ➡️ https://discord.com/channels/713503345364697088/1210336429855473705/1210706235033391186

there is outbound traffic though, looks at your graph very closely

inbound traffic would tend to generate outbound as your app answers the request

ah ok, indeed there are tiny variations in the outbound traffic

I'm assuming the inbound traffic is coming from the database connection since the public URL is disabled

although I just noticed that in the same project I have another API, not in Rails but in Node/Bun, that connects to a database which is a libSQL private instance inside the project and is also not sleeping, would the same apply even if the db for that one is not external/public?

private network traffic does not count towards inbound and outbound, in fact you can't even currently wake a service via the private network.

it must be something else then. The app in question is Node/Bun with these deps:

  "dependencies": {
    "@libsql/client": "0.5.6",
    "@paralleldrive/cuid2": "2.2.2",
    "bcrypt": "5.1.1",
    "drizzle-orm": "0.30.5",
    "elysia": "1.0.9",
    "jose": "5.2.3",
    "lodash": "4.17.21"
  },

perhaps something in there is doing telemetry or something

that's also possible

something seems off. The Bun app was sleeping before (I think… I don't have evidence other than the huge gap in the graph below). After an unintentional restart there's constant network traffic:

It restarted because I mistakenly disabled Private Networking for the Rails app without knowing that it was a project-wide setting, so all my apps deployed when I changed it

I changed it back (reenabled Private Networking for the project) and it re-deployed a second time, which resulted in that constant Network activity. I'll let it run overnight to see if this behavior continues

Updated 30d Network graph from above

could this Network traffic be investigated? There were no changes in the app code-wise. This refers to the Bun app that connects to the libSQL database via private networking. Logs don't show incoming requests for the past 2 days or so, so I'm at a loss here.

I now know that the Rails app will never go to sleep due to connecting to an external database, but that shouldn't be the case for the Bun app I think.

I'm sorry but this isn't something the team could help with

the team is only able to provide hobby users support for billing or platform issues

Then help me understand this graph a bit better so I can investigate myself on correct data:

• is private networking activity reported in that graph?

• is Docker-level (OS/Kernel/etc) activity (anything outside of my app code) reported there as well?

1. no

2. any traffic going outside and coming back of the private network is counted

so it's just real ingress/egress from my app? what if the Docker image for some reason uses an apt package that does telemetry for instance?

(I'm assuming it doesn't count)

that would be counted, since that would be outbound traffic

got it. Would that affect the app's ability to sleep? Assuming yes

yep, outbound traffic is what keeps the service awake

thanks for the answers. I'll try tweaking the Dockerfile as well as investigating traffic from the app, something like monkey-patching http.request in Node

feel free to close this (I don't know how)