#harbor-next/.dagger at build/daggerize ·...
echo prism
that's awesome news @plush tulip ! Where were the previous builds running on? public GHA?
plush tulip
That's awesome!!
Is the Hetzner machine long running?
Yes, we are running a couple of services there for CI/CD eg. Gitlab&GitHub runners, app dev instances, Cloud IDEs, and now Dagger.
Given the price of 47$ for 64Gib + on Intel i5-13500, this is the cheapest and most powerful instance we have
plush tulip
that's awesome news <@720375731015909437> ! Where were the previous builds runn...
different systems on runners from GitHub and our own GitLab runner on AWS.
time was similar +/- a couple of minutes.
now we are 10-20x faster.
most of the improvement happens from caching,
echo prism
most of the improvement happens from caching,
yes, good caching is a game changer. We're also making some fundamental changes and improvements on that area. You can expect more improvements to come
astral vigil
Gerhard shared some similar numbers with us, this is so cool to hear!
Main question for me is what does the long term maintenance/operational burden look like for fully self hosting.
Is there any other tooling you're using to help manage these or were you already running stuff in Hertzner outside of Dagger?
spiral swan
yes, good caching is a game changer. We're also making some fundamental changes ...
I'm super excited to see the caching changes. Is source code hashed and checked against the cache on the client side? I'd love to see something similar to the remote caching that Vercel and Turborepo offer, but don't know enough of the internals to know if that is possible