Everyone in this thread has gone absolutely insane. $5/month gets you 41 fucking _hours_ of continuous operation. If you're not utterly abusing the platform, this falls extremely below the threshold of caring. And if not, what the fuck are you even doing with all those hours? The new per-minute charge is less than one millisecond of engineer labor cost.

▲

seniorThrowaway 6 days ago | parent | next [-]

I have a nightly software build of a piece of software that takes 6 hours to create a 70GB artifact. The build process requires a GPU so it runs on my own HW. That's ~180 hours per month for this job alone. Is that really so hard to imagine?

▲

fbcpck 6 days ago | parent [-]

I don't know how much of that 6 hours build is tangled up in github workflows, but if it's a single contiguous block, you probably could make it near zero by making the self-hosted runner do only the preparation and only the final upload process (workflow_dispatch when the build is complete).

▲

seniorThrowaway 6 days ago | parent [-]

Most of it is just time waiting either while the source assets are downloaded (I clean slate it, that's the point of CI after all), the build itself runs, or the artifact is uploading to it's storage home. I'm sure it could be re-architected to use less actions minutes but if I'm going to redo it I will probably just move away from actions altogether because it's only loosely linked to Github anyway (runs on a schedule) and that way I am insulated from any future changes they come up with. The hardest part will likely be figuring out the Slack bot posting, I do use the marketplace action for that, but that's probably low lift. With LLM assisted coding I'm leaning more and more to little in house apps for stuff like this, it keeps you from dealing with lock in and other extractive gotchas.

▲

maratc 5 days ago | parent [-]

Slack posting is literally one-line curl with a token. That's what the fancy marketplace action does behind the scenes.

	▲	seniorThrowaway 5 days ago \| parent [-]
		Yep and a good example of how using the convenient best practice pushed by the vendor (the marketplace action) isn't a good idea. But I did.

▲

jrochkind1 6 days ago | parent | prev | next [-]

I think you significantly underestimate the number of CI minutes people are using in practice.

(Which, yes, has implications for energy use/climate change too for sure).

It doesn't look like i currently have access to the usage data on any of the lots-of-runners-lots-of-PRs projects I currently work on (which are still probably way less than some large companies).

	▲	BugsJustFindMe 6 days ago \| parent [-]
		> some large companies Any "large companies" don't give a shit about things at this cost level. They spend more on the time it takes you to open the door. The number of CI minutes could be astronomical and it still wouldn't rate above the threshold of caring. The time people in this thread have spent wringing their hands is way more expensive.

▲

fbcpck 6 days ago | parent | prev | next [-]

per minute billing is hard to wrap around the head

On my larger organization, we have on average 20 to 30 *active* runners during business hours. Assuming 5 on the off-hours, my napkin math says it comes down to about 10 fully-utilized-runners per month, so about 864$/mo. For the size of my organization that is honestly totally acceptable.

This is assuming 0.002$ per minute of job being actively executed. If it turns out to be 0.002$ per minute of *runner being registered* on the control plane, it would increase quite a bit. We are still using the old HorizontalRunnerAutoscaler with actions-runner-controller, with quite a pool of prewarmed runners idling to pick up a job. It would be a strong reason to use the new RunnerScaleSet (to take advantage of the reactive webhook-based scaling) and keep a very lean pool of prewarmed runners.

	▲	TheCondor 5 days ago \| parent [-]
		We have the same question, our runners are registers 24x7 but we probably only use a few hours a week. I get the logic of it, they have to have some sort of task running on their side when the runner is working. If it's only build time, then we don't care.

▲

JustFinishedBSG 6 days ago | parent | prev | next [-]

It could be 1 dollar a month, I'm still not paying to use my own ressources

	▲	jrochkind1 6 days ago \| parent [-]
		Well, you obviously are using their resources, to kick off and register statuses of the jobs running on your resources, right? That is probably worth $1/month to you?

▲

dap 6 days ago | parent | prev | next [-]

Doesn't this depend a lot on how long your actions run? Like, you may have already invested in your own hardware (maybe because your actions use a lot of resources and it's cheaper) and now you have to pay per-minute of action runtime for the API that does the bookkeeping?

▲

donatj 6 days ago | parent | prev [-]

$3 gets me 730 hours of comparable Vultr VPS time.