-
Notifications
You must be signed in to change notification settings - Fork 486
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Reducing downloads / decreasing build time #716
Comments
@ToeiRei feel your pain, ollama isn't a great benchmark indicative of most packages though, because most are using pip and those get cached on my pip server at This way has proven to scale for us across many packages, but some like ollama have different install processes that we try to remain compatible with as to avoid breaking changes. Ollama I typically have to update on a weekly or bi-weekly basis to keep it going... |
Yeah, it's a pain as I want to get a lot stuff built locally. Staged builds really help if you're able to keep the images and all the layers. Sadly my ssd isn't too big so I'm forced to purge. |
What I would recommend is to own a local copy of Artifact Manager as @dusty-nv is hosting at |
Hi guys,
I am building containers myself using
jetson-containers build
and it's "a tad slow" - so I decided to look at the build process; My 'benchmark image' is ollama as it builds relatively fast.My first hurdle is the huge apt repo pull which I solved by having a repo with a local apt-cacher. Speeds things a bit with the builds.
Looking at the Dockerfile and the dependency files, there are - in the case of ollama
among dependencies like cuda.
Is there a 'sane' way to cache or proxy those files?
As to my knowledge, squid won't help me there unless I intercept the HTTPS traffic (i.e. SSL bump) and thus get access to the decrypted content.
The text was updated successfully, but these errors were encountered: