Document Cluster Startup

[chrislovecnm]

1. addons are created in the s3 bucket by kops update cluster
2. nodeup creates the base manifests
3. nodeup downloads protokube from s3 in a tarball and is imported into docker via docker import
4. protokube mounts etcd disks
5. protokube bootstraps etcd
6. protokube sets etcd DNS endpoints
7. kubectl starts, everything is in eventual consistency
8. etcd starts
9. API server comes up

then we have a base bootstrapped cluster
everything is starting

protokube has two loops in it

1. it makes sure kubelet is running (important)
2. it executes channels

Channels is a binary that has the 'addon' api

1. it parses the addons from the addons folder in the state store
2. and runs kubectl apply

anything that is not a core k8s component or etcd is an addon

CNI
dns-controller
kube-dns

DNS Controller

1. one of the first deployments to start on the master
2. it sets the dns for API of the master

[wlan0]

Thanks for adding this issue @chrislovecnm

In addition, could we also document how to debug addons startup by the channels binary?

I understand that addon manifests are available in ${STATE_STORE}/${CLUSTER_NAME}/addons

I think it'll be good to document where to find channels logs, and also how to find if there are any errors during addon startup

[chrislovecnm]

Channels binary runs inside of protokube, but you can run it locally :) Just take a look at the protokube logs on a master.

kops update cluster <- builds the manifests and uploads them to s3. They are k8s manifests. That is when boostrapchannelbuilder.go runs.
protokube executes channels
channels reads the addons yaml file, and executes kubectl
kubectl apply my-addon.yaml <- that is weave, kube-dns etc The manifest lives in s3

[fejta-bot]

Issues go stale after 90d of inactivity.
Mark the issue as fresh with /remove-lifecycle stale.
Stale issues rot after an additional 30d of inactivity and eventually close.

Prevent issues from auto-closing with an /lifecycle frozen comment.

If this issue is safe to close now please do so with /close.

Send feedback to sig-testing, kubernetes/test-infra and/or @fejta.
/lifecycle stale

[fejta-bot]

Stale issues rot after 30d of inactivity.
Mark the issue as fresh with /remove-lifecycle rotten.
Rotten issues close after an additional 30d of inactivity.

If this issue is safe to close now please do so with /close.

Send feedback to sig-testing, kubernetes/test-infra and/or fejta.
/lifecycle rotten
/remove-lifecycle stale

[chrislovecnm]

/lifecycle frozen
/remove-lifecycle stale