docs: clarify kind prereqs and fix stale install-script references#32
docs: clarify kind prereqs and fix stale install-script references#32Davanum Srinivas (dims) wants to merge 1 commit into
Conversation
Davanum Srinivas (dims)
force-pushed
the
docs/kind-local-dev-prereqs
branch
from
ceef546 to
97ff0d1
Compare
- README Quickstart now pins kind >= v0.31.0. Earlier releases ship a default Kubernetes node image without the PodCertificateRequest API the install scripts depend on, so kube-apiserver never reaches readiness on a fresh `brew install kind` machine. - README Quickstart also documents the 12 GiB Docker Desktop memory floor required for the full system bring-up. - internal/e2e/readme.md referenced a `hack/install.sh` that does not exist. Point readers at the two real entry points instead (hack/install-ate.sh for GKE, hack/install-ate-kind.sh for kind). - demos/counter/README.md claimed a GCS bucket was required, but the kind overlay provisions an in-cluster rustfs S3 store. Document both targets so the demo can run without provisioning GCP. Signed-off-by: Davanum Srinivas <davanum@gmail.com>
Davanum Srinivas (dims)
force-pushed
the
docs/kind-local-dev-prereqs
branch
from
97ff0d1 to
6f2fae0
Compare
|
Thanks, one thing: I'd like to eliminate those pre-reqs as much as possible #46 |
Benjamin Elder (BenTheElder)
added
the
documentation
|
|
||
| 1. Make sure you have [`kind`](https://kind.sigs.k8s.io/), [`kubectl`](https://kubernetes.io/docs/tasks/tools/), and [`docker`](https://www.docker.com/) installed and configured on your dev machine. | ||
| 1. Make sure the following tools are installed and configured on your development machine: | ||
| - [`kind`](https://kind.sigs.k8s.io/) **v0.31.0 or newer** (required for the `PodCertificateRequest` API the install scripts use; matches the CI-pinned `go install sigs.k8s.io/kind@v0.31.0`). |
There was a problem hiding this comment.
not required or used anymore
There was a problem hiding this comment.
Benjamin Elder (@BenTheElder) we still use (
)should we reflect that? (and drop the note about PodCertificateRequest API)
There was a problem hiding this comment.
If you DIY install we don't inherently depend on kind, but our dev scripts will depend on the version in the tools module in the future, I think we don't need to specify the version here. Thanks.
I also think we will stop depending on the default node image in the future, but I want to ship some upstream kind fixes first.
| 1. Make sure the following tools are installed and configured on your development machine: | ||
| - [`kind`](https://kind.sigs.k8s.io/) **v0.31.0 or newer** (required for the `PodCertificateRequest` API the install scripts use; matches the CI-pinned `go install sigs.k8s.io/kind@v0.31.0`). | ||
| - [`kubectl`](https://kubernetes.io/docs/tasks/tools/). | ||
| - [`docker`](https://www.docker.com/) with **at least 12 GiB of memory** allocated to the VM (Docker Desktop: Settings → Resources → Memory). |
There was a problem hiding this comment.
are we using 12GiB?
must be the image cache ... we're going to need to think about tiers there
There was a problem hiding this comment.
12 seemed to work :) i did not try going lower
There was a problem hiding this comment.
Hmm OK, I want to avoid overstating it if we can, this is also likely going to be very fluid as we iterate, most of it is probably various in-memory storage currently that doesn't strictly need to all be in memory in the future ...
2 participants
brew install kindmachine.hack/install.shthat does not exist. Point readers at the two real entry points instead (hack/install-ate.sh for GKE, hack/install-ate-kind.sh for kind).Fixes #<issue_number_goes_here>