Hello!
I’ve been experimenting with Fly, and a few weeks ago I changed an app (namely, Pomerium) from running in a “scale to zero” configuration to always running a single machine. A few days later, I happened to notice the “Outbound Bandwidth” row on the Usage dashboard shoot up to about 5GB per day (+1800%!), and so I dug in.
The short answer is that the outbound traffic is due to metrics scrapes - when I have metrics enabled for Pomerium, I notice that “Outbound Bandwidth: North America and Europe” on the Billing dashboard consistently ticks up by a few cents per day - but when I disable metrics, it goes up by maybe a cent every day or so.
When I hit Pomerium’s metrics endpoint via wget
, I get something like an 800K response back - that’s a fairly heavy scrape, but I’m mostly wondering if outbound traffic due to metrics scrapes is supposed to incur billing or not!