Do POST requests restart stopped machines?

1

:warning: Potentially a XY problem question, but hopefully I can have some insights.

I have two hobby Telegram bots deployed to Fly.io (if that matters, Airnope to kill cryptocurrency spa, and a simple list app, Bot na Lista), both written in Rust with Teloxide.

I had to set them both to:

  auto_stop_machines = false
  min_machines_running = 1

So they don’t suspend/stop the machines. I wasn’t expecting to need that, because the way Telegram bots work is via webhook, meaning, every time there is a message for the Bot, Telegram servers send a POST request to Fly.io (to my bot).

I expected that to re-start the suspended or stop machines, but it does not. Is that expected?

Any ideas to debug the whole integration?

Many many thanks,

2

I think any http request would wake the machine. Your web hooks are failing likely because of the latency timeout from waking up.

3

Your web hooks are failing likely because of the latency timeout from waking up.

I don’t think this is that likely. Telegram re-attempts the webhook if it fails:

In case of an unsuccessful request, we will give up after a reasonable amount of attempts.

So what I expected was a slow response, even a few minutes (maybe 1h). But nope. The bot only wakes up is I manually re-start the machine. Weird, eh?

4

Maybe the issue is related to the degraded API in the past 24 hours.

6

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.