There was an absurd amount of RAM billed recently from my account, and I can’t seem to find where it is coming from.
What kind of dark magic did I do to get this dimension of additional RAM?
I’m serving a simple app with Go in the backend and Angular in the frontend.
I would like to understand the 1279800 GB/s better. Is it how much I consumed of extra RAM? If not, what is it? If that’s it, what could be the cause?
Just so we’re on the same page, are you reading “1279800 GB/s” as “1279800 Gigabyte-seconds”, or " 1279800 Gigabytes per second"? It should be the former, not the latter. You have four Machines each with 1 Gigabyte of memory, so with 4 Gigabytes of memory total, 1279800 Gigabyte-seconds is about 4 days.
Hi, sorry this is confusing at the moment, we’re working on making this easier to understand.
Additional RAM is explained in our pricing docs. I can see from the invoice that you’re using only shared machines.
If, for example, you’re using shared-cpu-1x 512mb, then 256mb of RAM is “additional” to the amount given in the named shared-cpu-1x preset (which includes 256mb).
If you’re using shared-cpu-1x 1024mb, then 768mb is additional. This will cost 768/1024 * $5 (number of GB * roughly the cost per GB per month of additional RAM) = $3.75 per month per machine, if the machine is running for the whole month. Stopped machines do not incur any additional RAM charge (only rootfs charges).
Oh, that makes a lot more sense. I was reading as the former, thanks for clarifying.
I see. Although, from what I can get, most of this additional RAM is being used by builders. If I build the image and send it to the registry directly instead, should I expect that this bill drops considerably?
Builders don’t cost you anything, they don’t show up in usage at all. We only record usage (and bill) for apps/machines that you created.
To be super clear: the additional RAM charges you see are due to machines you’ve created.
If you want to reduce the amount you’re being charged for additional RAM, you will need to scale the amount of RAM you’ve configured for your machines (and check your fly.toml to ensure that a subsequent deploy won’t undo any such changes).
FWIW “GB/s” does mean Gigabytes-per-second.
It seems like it should be Gigabyte-seconds as you said, so the unit should be GBs or GB-s. It’s probably worth updating the invoices to reflect this.
In the case of the OP, 1,279,800 GB-s is the equivalent of using 1GB of memory for ~14 days (86400 seconds in a day); or 4GB for ~3.7days
Ok, thanks for the info, I will look into the scaling of RAM.
This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.