1# Edge Clusterctl Pallet
2
3Since this pallet contains GCP infra, split scheduling is recommended. Read the documentation on [Applying Pallet Packages](https://docs.edge-infra.dev/foundation/manifest-scheduling/pallets/developing-pkgs/applying/) to learn about split scheduling and its prerequisites. Also, getting/having access to a GCP development sandbox is required.
4
5## Environment Setup
6
7Create an .env file with the following variables:
8```dotenv
9# gke, sds, or generic. use "generic" for any K8s cluster that isn't GKE or
10# SDS.
11export CLUSTER_PROVIDER=
12# uuid is necessary for generating a cluster_hash, existing resources should
13# have this in their annotations already but one can be created by using
14# uuidgen
15export CLUSTER_UUID=
16# google cloud project id
17export GCP_PROJECT_ID=
18# defaults to us and east1-b, otherwise needs to be valid GCP value
19export GCP_REGION=
20export GCP_ZONE=
21# set to same value as gcp project ID when developing within a single
22# project.
23export FOREMAN_GCP_PROJECT_ID=
24# enables split scheduling
25export INFRA=true
26# name of the kubeconfig context to use to schedule infra objects;
27# default behavior is to apply all objects to the same K8s context
28export INFRA_CONTEXT=
29# target namespace for all applied infrastructure resources, including
30# dependencies; the namespace is created if it does not already exist.
31export INFRA_NAMESPACE=
32```
33
34## Applying the Pallet
35
36The `clusterctl` pallet requires the `edge-backend-launch-darkly-sdk-key` and `edge-backend-totp-secret` to exist on the cluster it is being applied to. Check out the [Create Secret Manager Secrets](https://docs.edge-infra.dev/edge/operations/bootstrapping/#create-secret-manager-secrets) section of the Bootstrapping docs for more information.
37
38Tips for applying:
39- The `key.json` can be created by going to the GCP console for your sandbox project, finding the IAM service account for kcc and adding a key in the `key` section of the service account
40- If you are using a `kind` cluster, you can reset it with:
41```
42just hack/kind/reset
43```
44Then, run these commands on the non-infra cluster:
45```shell
46source .env
47
48lift apply \
49 --domain= \
50 --env-folder-id= \
51 config/pallets/edge/clusterctl
52```
53
54*Note:* Some IAM policy members may not have the permissions required to be configured correctly depending on where the infra for the pallet is applied. The `deployment` should still run even if this issue arises
55
56> If an error similar to this is returned: `dry-run failed, error: failed to create typed patch object: .data.ENV_FOLDER_ID: expected string, got &value.valueUnstructured{Value:<your_id>}`, then wrap the env-folder-id value with double & single quotes (e.g. '"0123456789"') and apply the pallet again
57
58## Rendering Parameters
59
60Check the `clusterctl`'s IAM policy members in `config/pallets/edge/clusterctl/gcpinfra/manifests.yaml` for more context on where most of these rendering parameters are being used. All gcp id's were taken from the GCP console (click select a project -> click `ALL` tab -> expand the `preprod` folder -> explore this directory).
61
62| Rendering Parameter | Description | Example |
63|---------------------|-----------------------------------------------------|---------------------|
64| domain | domain id for the instance | edge-preprod.dev |
65| env-folder-id | gcp folder id for the instance's environment folder | 869642396943 (dev0) |
View as plain text