Making a different app machine my free builder

[Ethan888]

I currently have a machine running under a personal plan. This machine is only allocated 228 MB - where as I have another machine (that I do not / never used), that has the 4GB “Free Builder” plan.

I am wondering if I can somehow apply the Free Builder plan to the machine I actually use?

Edit: I deleted the Free Builder plan machine because I never used it. I now no longer have access to a free builder plan.

[ziding]

I believe the free builder is dedicated to building your docker images, not hosting the app itself.

I also believe a new free builder is provisioned so long as you don’t deploy with --local-only.

[kaz]

@Ethan888

I am wondering if I can somehow apply the Free Builder plan to the machine I actually use?

Edit: I deleted the Free Builder plan machine because I never used it. I now no longer have access to a free builder plan.

No. Free Builder machines are for building container images and we are assuming customers are not building images 24/7. We have some guardrails that prevent customers to use the builder machines for something else.

Regarding the lack of it, fly deploy should provision a new one if you don’t have it (or you removed it).

[Ethan888]

Hey Kaz, thank you for the quick reply. That clears up my confusion (for the most part). That said, I am still new to hosting on machines. May I ask you some questions?

Background: I am currently running a discord bot on a free-plan machine. Responses from OpenAI streaming are… slow. Upwards of 30 seconds from the time a message is sent to a response from OpenAI. I assume this is because Openai needs to wait for the message to be streamed in letter-by-letter, but it could also be the server’s capabilities resulting in long response times.

1. Would upgrading to a paid VM help speed up my responses (enough to be noticeable), or is this a limitation by openai’s text streaming?

2. For when I do upgrade my plan, do you recommend any specific plan/machine/etc. that can help my discord bot’s performance?

Thank you again for your support.

Edit: Here is a picture showing my current usage

[kaz]

I do think OpenAI is the bottleneck here, not Fly.io. That said, you can scale up your machines to confirm whether the machines are the bottleneck or not.