Commit c4f842f1 authored by Daniel Gruesso's avatar Daniel Gruesso

Merge branch 'master' into 'master'

Documentation for using Crossplane as a managed app

See merge request gitlab-org/gitlab!19675
parents f4a467ab db046d44
......@@ -40,6 +40,7 @@ The following applications can be installed:
- [GitLab Runner](#gitlab-runner)
- [JupyterHub](#jupyterhub)
- [Knative](#knative)
- [Crossplane](#crossplane)
With the exception of Knative, the applications will be installed in a dedicated
namespace called `gitlab-managed-apps`.
......@@ -384,6 +385,38 @@ chart is used to install this application with a
[`values.yaml`](https://gitlab.com/gitlab-org/gitlab/blob/master/vendor/prometheus/values.yaml)
file.
### Crossplane
> - [Introduced](https://gitlab.com/gitlab-org/gitlab/issues/34702) in GitLab 12.5 for project-level clusters.
[Crossplane](https://crossplane.io/docs) is a multi-cloud control plane useful for
managing applications and infrastructure across multiple clouds. It extends the
Kubernetes API using:
- Custom resources.
- Controllers that watch those custom resources.
Crossplane allows provisioning and lifecycle management of infrastructure components
across cloud providers in a uniform manner by abstracting cloud provider-specific
configurations.
The Crossplane GitLab-managed application:
- Installs Crossplane with a provider of choice on a Kubernetes cluster attached to the
project repository.
- Can then be used to provision infrastructure or managed applications such as
PostgreSQL (for example, CloudSQL from GCP or RDS from AWS) and other services
required by the application via the Auto DevOps pipeline.
For information on configuring Crossplane installed on the cluster, see
[Crossplane configuration](crossplane.md).
NOTE: **Note:**
[`alpha/crossplane`](https://charts.crossplane.io/alpha/) chart v0.4.1 is used to
install Crossplane using the
[`values.yaml`](https://github.com/crossplaneio/crossplane/blob/master/cluster/charts/crossplane/values.yaml.tmpl)
file.
## Upgrading applications
> [Introduced](https://gitlab.com/gitlab-org/gitlab-foss/merge_requests/24789) in GitLab 11.8.
......@@ -424,6 +457,7 @@ The applications below can be uninstalled.
| JupyterHub | 12.1+ | All data not committed to GitLab will be deleted and cannot be restored. |
| Knative | 12.1+ | The associated IP will be deleted and cannot be restored. |
| Prometheus | 11.11+ | All data will be deleted and cannot be restored. |
| Crossplane | 12.5+ | All data will be deleted and cannot be restored. |
To uninstall an application:
......
# Crossplane configuration
Once Crossplane [is installed](applications.md#crossplane), it must be configured for
use.
The process of configuring Crossplane includes:
1. Configuring RBAC permissions.
1. Configuring Crossplane with a cloud provider.
1. Configure managed service access.
1. Setting up Resource classes.
1. Using Auto DevOps configuration options.
1. Connect to the PostgreSQL instance.
To allow Crossplane to provision cloud services such as PostgreSQL, the cloud provider
stack must be configured with a user account. For example:
- A service account for GCP.
- An IAM user for AWS.
Important notes:
- This guide uses GCP as an example. However, the process for AWS and Azure will be
similar.
- Crossplane requires the Kubernetes cluster to be VPC native with Alias IPs enabled so
that the IP address of the pods are routable within the GCP network.
First, we need to declare some environment variables with configuration that will be used throughout this guide:
```sh
export PROJECT_ID=crossplane-playground # the GCP project where all resources reside.
export NETWORK_NAME=default # the GCP network where your GKE is provisioned.
export REGION=us-central1 # the GCP region where the GKE cluster is provisioned.
```
## Configure RBAC permissions
- For a non-GitLab managed cluster(s), ensure that the service account for the token provided can manage resources in the `database.crossplane.io` API group.
Manually grant GitLab's service account the ability to manage resources in the
`database.crossplane.io` API group. The Aggregated ClusterRole allows us to do that.
NOTE: **Note:**
For a non-GitLab managed cluster, ensure that the service account for the token provided can manage resources in the `database.crossplane.io` API group.
​1. Save the following YAML as `crossplane-database-role.yaml`:
```sh
cat > crossplane-database-role.yaml <<EOF
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRole
metadata:
name: crossplane-database-role
labels:
rbac.authorization.k8s.io/aggregate-to-edit: "true"
rules:
- apiGroups:
- database.crossplane.io
resources:
- postgresqlinstances
verbs:
- get
- list
- create
- update
- delete
- patch
- watch
EOF
```
Once the file is created, apply it with the following command in order to create the necessary role:
```sh
kubectl apply -f crossplane-database-role.yaml
```
## Configure Crossplane with a cloud provider
See [Configure Your Cloud Provider Account](https://crossplane.io/docs/v0.4/cloud-providers.html)
to configure the installed cloud provider stack with a user account.
Note that the Secret and the Provider resource referencing the Secret needs to be
applied to the `gitlab-managed-apps` namespace in the guide. Make sure you change that
while following the process.
[Configure Providers](https://crossplane.io/docs/v0.4/cloud-providers.html)
## Configure Managed Service Access
We need to configure connectivity between the PostgreSQL database and the GKE cluster.
This can done by either:
- Using Crossplane as demonstrated below.
- Directly in the GCP console by
[configuring private services access](https://cloud.google.com/vpc/docs/configure-private-services-access).
Create a GlobalAddress and Connection resources:
```sh
cat > network.yaml <<EOF
---
# gitlab-ad-globaladdress defines the IP range that will be allocated for cloud services connecting to the instances in the given Network.
apiVersion: compute.gcp.crossplane.io/v1alpha3
kind: GlobalAddress
metadata:
name: gitlab-ad-globaladdress
spec:
providerRef:
name: gcp-provider
reclaimPolicy: Delete
name: gitlab-ad-globaladdress
purpose: VPC_PEERING
addressType: INTERNAL
prefixLength: 16
network: projects/$PROJECT_ID/global/networks/$NETWORK_NAME
---
# gitlab-ad-connection is what allows cloud services to use the allocated GlobalAddress for communication. Behind
# the scenes, it creates a VPC peering to the network that those service instances actually live.
apiVersion: servicenetworking.gcp.crossplane.io/v1alpha3
kind: Connection
metadata:
name: gitlab-ad-connection
spec:
providerRef:
name: gcp-provider
reclaimPolicy: Delete
parent: services/servicenetworking.googleapis.com
network: projects/$PROJECT_ID/global/networks/$NETWORK_NAME
reservedPeeringRangeRefs:
- name: gitlab-ad-globaladdress
EOF
```
Apply the settings specified in the file with the following command:
```sh
kubectl apply -f network.yaml
```
You can verify creation of the network resources with the following commands.
Verify that the status of both of these resources is ready and is synced.
```sh
kubectl describe connection.servicenetworking.gcp.crossplane.io gitlab-ad-connection
kubectl describe globaladdress.compute.gcp.crossplane.io gitlab-ad-globaladdress
```
## Setting up Resource classes
Resource classes are a way of defining a configuration for the required managed service. We will define the Postgres Resource class
- Define a gcp-postgres-standard.yaml resourceclass which contains
1. A default CloudSQLInstanceClass.
1. A CloudSQLInstanceClass with labels.
```sh
cat > gcp-postgres-standard.yaml <<EOF
apiVersion: database.gcp.crossplane.io/v1beta1
kind: CloudSQLInstanceClass
metadata:
name: cloudsqlinstancepostgresql-standard
labels:
gitlab-ad-demo: "true"
specTemplate:
writeConnectionSecretsToNamespace: gitlab-managed-apps
forProvider:
databaseVersion: POSTGRES_9_6
region: $REGION
settings:
tier: db-custom-1-3840
dataDiskType: PD_SSD
dataDiskSizeGb: 10
ipConfiguration:
privateNetwork: projects/$PROJECT_ID/global/networks/$NETWORK_NAME
# this should match the name of the provider created in the above step
providerRef:
name: gcp-provider
reclaimPolicy: Delete
---
apiVersion: database.gcp.crossplane.io/v1beta1
kind: CloudSQLInstanceClass
metadata:
name: cloudsqlinstancepostgresql-standard-default
annotations:
resourceclass.crossplane.io/is-default-class: "true"
specTemplate:
writeConnectionSecretsToNamespace: gitlab-managed-apps
forProvider:
databaseVersion: POSTGRES_9_6
region: $REGION
settings:
tier: db-custom-1-3840
dataDiskType: PD_SSD
dataDiskSizeGb: 10
ipConfiguration:
privateNetwork: projects/$PROJECT_ID/global/networks/$NETWORK_NAME
# this should match the name of the provider created in the above step
providerRef:
name: gcp-provider
reclaimPolicy: Delete
EOF
```
Apply the resource class configuration with the following command:
```sh
kubectl apply -f gcp-postgres-standard.yaml
```
Verify creation of the Resource class with the following command:
```sh
kubectl get cloudsqlinstanceclasses
```
The Resource Classes allow you to define classes of service for a managed service. We could create another `CloudSQLInstanceClass` which requests for a larger or a faster disk. It could also request for a specific version of the database.
## Auto DevOps Configuration Options
The Auto DevOps pipeline can be run with the following options:
The Environment variables, `AUTO_DEVOPS_POSTGRES_MANAGED` and `AUTO_DEVOPS_POSTGRES_MANAGED_CLASS_SELECTOR` need to be set to provision PostgresQL using Crossplane
Alertnatively, the following options can be overridden from the values for the helm chart.
- `postgres.managed` set to true which will select a default resource class.
The resource class needs to be marked with the annotation
`resourceclass.crossplane.io/is-default-class: "true"`. The CloudSQLInstanceClass
`cloudsqlinstancepostgresql-standard-default` will be used to satisfy the claim.
- `postgres.managed` set to `true` with `postgres.managedClassSelector`
providing the resource class to choose based on labels. In this case, the
value of `postgres.managedClassSelector.matchLabels.gitlab-ad-demo="true"`
will select the CloudSQLInstance class `cloudsqlinstancepostgresql-standard`
to satisfy the claim request.
The Auto DevOps pipeline should provision a PostgresqlInstance when it runs succesfully.
Verify creation of the PostgresQL Instance.
```sh
kubectl get postgresqlinstance
```
Sample Output: The `STATUS` field of the PostgresqlInstance transitions to `BOUND` when it is successfully provisioned.
```
NAME STATUS CLASS-KIND CLASS-NAME RESOURCE-KIND RESOURCE-NAME AGE
staging-test8 Bound CloudSQLInstanceClass cloudsqlinstancepostgresql-standard CloudSQLInstance xp-ad-demo-24-staging-staging-test8-jj55c 9m
```
The endpoint of the PostgreSQL instance, and the user credentials, are present in a secret called `app-postgres` within the same project namespace.
Verify the secret with the database information is created with the following command:
```sh
kubectl describe secret app-postgres
```
Sample Output:
```
Name: app-postgres
Namespace: xp-ad-demo-24-staging
Labels: <none>
Annotations: crossplane.io/propagate-from-name: 108e460e-06c7-11ea-b907-42010a8000bd
crossplane.io/propagate-from-namespace: gitlab-managed-apps
crossplane.io/propagate-from-uid: 10c79605-06c7-11ea-b907-42010a8000bd
Type: Opaque
Data
====
privateIP: 8 bytes
publicIP: 13 bytes
serverCACertificateCert: 1272 bytes
serverCACertificateCertSerialNumber: 1 bytes
serverCACertificateCreateTime: 24 bytes
serverCACertificateExpirationTime: 24 bytes
username: 8 bytes
endpoint: 8 bytes
password: 27 bytes
serverCACertificateCommonName: 98 bytes
serverCACertificateInstance: 41 bytes
serverCACertificateSha1Fingerprint: 40 bytes
```
## Connect to the PostgresQL instance
Follow this [GCP guide](https://cloud.google.com/sql/docs/postgres/connect-kubernetes-engine) if you
would like to connect to the newly provisioned Postgres database instance on CloudSQL.
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment