Hi, I’m curious about the Fly Machine pricing model. For a case whereby I’m running user code on it, would it cost the same when the machines are turned off? I couldn’t understand it from the pricing page. Ideally, I’m hoping for something where I don’t have to pay for a machine that’s turned off.
I saw some reviews when it was launched and it seemed you charge for machines that are turned off, but you were reviewing that option. I’m not sure if anything has changed in that regard
If your machine is off, you won’t be billed for compute, i.e. CPU seconds. You’re still billed for a volume attached to a stopped machine.
We’re also soon going to start billing for rootfs of stopped machines. We said we’d bill for this when we released machines (last section of this blog post), but never implemented it at the time. We’re working on it now, but we’ll give plenty of notice before we actually start charging.
Hi… Do you mean the root partition (where / is mounted) or the Docker image?
(The link only mentions the latter.)
Yeh the root partition. But to clarify, this has the same size as the app’s docker image, since we’re only going to be charging for the “base” size.
In other words, if your machine starts and then downloads a bunch of stuff into /, you won’t get billed for that stuff when you stop the machine as part of the rootfs usage, because we don’t keep hold of it.
The reasoning behind this is that in order to have quick cold starts, we have to keep the built root partition on a host. Otherwise every cold start would involve pulling from the registry, which is much slower. So there’s a non-zero resource cost to quick cold starts.
There’ll be a fresh produce post about this with more details at some point, if the above is unclear.
This topic was automatically closed 2 days after the last reply. New replies are no longer allowed.