Machines Keep Stopping with "Excess Capacity"

[sinanspd]

I am trying to get familiar with fly.io so I deployed a Shopify Remix app today to try to figure out the basics. One thing I noticed is that the both machines hosting the app keep stopping every few minutes with logs saying App kronos-vip-on-site has excess capacity, auto stopping machine 784ee59fd20198. 0 out of 1 machines left running (region=dfw, process group=app), followed by a SIGINT being sent to the app process and Virtual machine exited abruptly. There are no errors being logged prior to the shutdown.

The machine comes back up successfully when the app is used (refreshed page etc.). I am on the free plan currently but I checked all the metrics and there is no resource coming close to any of the limits (a few Kbs of data in and out, the DB has 2 rows etc.) I am not entirely sure what “excess capacity” is referring to but it definitely can’t be compute resources. This isn’t a public app yet. I am just testing the waters myself. I read on Shopify’s docs that Fly sleeps containers (thus the need for volumes for storage etc.) and I do have auto_stop_machines = true but these shutdowns seem way too frequent.

Is this expected behavior for an idle app? Or is something else going on?

[andie]

hi @sinanspd

You are seeing the expected behaviour of Fly Proxy autostopping your Machines when there’s no (or less) traffic to your app. There’s an explanation of how all this works here: Automatically stop and start Machines · Fly Docs. But the important bit for your question is that the process to check whether an app has excess capacity happens every 2 minutes or so (or 5 minutes after a new deploy) which is probably the frequency you’re seeing in logs. And “excess capacity” just means that your app isn’t using all the running resources and Machines can be shut down until they are needed again.