Deployment

Prerequisites
Build Images
Deploy Database
Run Db migration
Deploy API Service
Deploy UI
Update Catalog and Catalog Refresh
Schedule Catalog Refresh On an Interval
Setup Catalog Refresh CronJob
Adding New Users in Config
Deploying on Disconnected Cluster
Troubleshooting

Prerequisites

Kubernetes Cluster
- You can also deploy on Minkube, kind or OpenShift.
kubectl
docker or podman

Build Images

Lets build and push all images to a registry and then create deployments.

Note : You can use images from quay and skip the build image section.

The image tag would be same as release tag. You can check the latest release here
For example:
- UI Image: quay.io/tekton-hub/ui:v1.3.0
- DB Migration Image: quay.io/tekton-hub/db-migration:v1.3.0
- API Image: quay.io/tekton-hub/api:v1.3.0
To Build images for multi-architecture use the following command and default value of architecture is linux/amd64.

docker buildx build  --platform linux/amd64,linux/s390x,linux/ppc64le -f images/api.Dockerfile -t <image> --push .

UI Image

Ensure you are in the root of project and logged in you image registry.

docker buildx build -f images/ui.Dockerfile -t <image> --push .

Replace <image> with the registry and image name. eg. quay.io/<username>/ui

API & DB Migration Image

Ensure you are in the root of project and logged in you image registry.

Build the API image using below command

docker buildx build -f images/api.Dockerfile -t <image> --push .

Replace <image> with the registry and image name. eg. quay.io/<username>/api

Now, Build the Db migration image using below command

docker buildx build -f images/db.Dockerfile -t <image> --push .

Replace <image> with the registry and image name. eg. quay.io/<username>/db-migration

Make sure all images are public before creating deployments.

NOTE: If You have deployed Hub v1.6.0 and want to upgrade to Hub v1.7.0 then please follow this doc for next steps. As we have renamed config metadata name and labels in the configs of db, api and ui.

NOTE: In case of using Github or Gitlab Enterprise make sure you deploy Hub in the same network where the Enterprise server is running. Example if your Github/Gitlab Enterprise is running behind VPN then your Kubernetes Cluster should also be behind VPN.

Deploy Database

Now, we have all images pushed to registry. Lets deploy the database first.

Navigate to config directoy in the root of the project.

Execute below command

kubectl apply -f 00-init/ -n tekton-hub

This will create tekton-hub namespace, db deployment with postgres image, secret to save db credentials, pvc for the db and a ClusterIP service to expose the db internally.

All resources are created in tekton-hub namespace.

Wait till the pod comes in a running state

kubectl get pod -n tekton-hub -w

Run DB migration

Once the pod is in running state now we can run db migration. This will create all the tables required in the database.

Edit the 01-db/10-db-migration.yaml and replace the image you created previously and apply the yaml.

kubectl apply -f 01-db/10-db-migration.yaml -n tekton-hub

This will create a job which will read the db credentials from the secret created while deploying database and create all required tables.

$ kubectl get pods -n tekton-hub
NAME                   READY   STATUS      RESTARTS   AGE
tekton-hub-db-589d44fdd5-ksf8v    1/1     Running     0          50s
tekton-hub-db-migration-8vhpd     0/1     Completed   0          17s

The Pod will go into Completed state if the job succeeds. You can also check the logs using kubectl logs -n tekton-hub tekton-hub-db-migration-8vhpd. In case of any failure the Pod will go into Error state.

$ kubectl get pods -n tekton-hub
NAME                   READY   STATUS      RESTARTS   AGE
tekton-hub-db-589d44fdd5-ksf8v    1/1     Running     0          50s
tekton-hub-db-migration-8vhpd     0/1     Error       0          17s

Deploy API Service

Setup Route or Ingress

If deploying on OpenShift:-

kubectl apply -f 04-openshift/40-api-route.yaml -f 04-openshift/40-auth-route.yaml -n tekton-hub

If deploying on Kubernetes:-

Create the secret containing tls cert and tls key both for API and Oauth server. Example as follows:

kubectl create secret tls api-hub-tekton-dev-tls --cert=path/to/cert/file --key=path/to/key/file -n tekton-hub

Apply the Ingress

kubectl apply -f 04-kubernetes/40-api-ingress.yaml -f 04-kubernetes/40-auth-ingress.yaml -n tekton-hub

Create Git Oauth Applications

Github - Create a GitHub OAuth with Homepage URL and Authorization callback URL as <auth-route>. Follow the steps given here to create a GitHub OAuth.
Gitlab - Create a Gitlab Oauth with REDIRECT_URI as <auth-route>/auth/gitlab/callback. Follow the steps given here to create a Gitlab Oauth
BitBucket - Create a BitBucket Oauth with Callback URL as <auth-route>. Follow the steps given here to create a BitBucket Oauth

Update API Secret

Edit 02-api/20-api-secret.yaml and update the configuration

After creating the OAuth add the Client ID and Client Secret in the yaml file.
For JWT_SIGNING_KEY, you can add any random string, this is used to sign the JWT created for users.
For ACCESS_JWT_EXPIRES_IN and REFRESH_JWT_EXPIRES_IN add time you want the jwt to be expired in. Refresh time should be greater than Access time. eg. 1m = 1 minute, 1h = 1 hour, 1d = 1 day
- NOTE: Supported formats for ACCESS_JWT_EXPIRES_IN and REFRESH_JWT_EXPIRES_IN are w(weeks), d(days), h(hours), m(min) and s(sec)

apiVersion: v1
kind: Secret
metadata:
  name: tekton-hub-api
  namespace: tekton-hub
type: Opaque
stringData:
  GH_CLIENT_ID: Github Oauth client id
  GH_CLIENT_SECRET: Github Oauth secret
  GL_CLIENT_ID: Gitlab Oauth client id
  GL_CLIENT_SECRET: Gitlab Oauth secret
  BB_CLIENT_ID: BitBucket Oauth client id
  BB_CLIENT_SECRET: BitBucket Oauth secret
  JWT_SIGNING_KEY: a-long-signing-key
  ACCESS_JWT_EXPIRES_IN: time such as 15m
  REFRESH_JWT_EXPIRES_IN: time such as 15m
  AUTH_BASE_URL: auth route
  GHE_URL: Add Github Enterprise URL in case of authenticating through Github Enterprise (Example (https|http)://myghe.com) --> Do not provide the catalog URL
  GLE_URL: Add Gitlab Enterprise URL in case of authenticating through Gitlab Enterprise (Example (https|http)://mygle.com) --> Do not provide the catalog URL

Update API ConfigMap

ConfigMap (02-api/21-api-configmap.yaml) contains information for Api such as categories, default Catalog tektoncd and Scopes.
This configMap have the enough information to deploy the default catalog tektoncd, if changes are required then the ConfigMap should be modified accordingly .
Update the ConfigMap to add at least one user with all scopes such as
For example
- refresh a catalog,
- refresh config file
- create an agent token

scopes:
  - name: agent:create
    users: [foo]        <<< Where `foo` is your Github Handle
  - name: catalog:refresh
    users: [foo]
  - name: config:refresh
    users: [foo]

Final ConfigMap would look like below:

apiVersion: v1
kind: ConfigMap
metadata:
  name: tekton-hub-api
  labels:
    app: tekton-hub-api
data:
  CATALOG_REFRESH_INTERVAL: 30m ## change the catalog refresh interval here
  CATEGORIES: |
    - Automation
    - Build Tools
  CATALOGS: | ## Add the new catalog information here
    - name: tekton
      org: tektoncd
      type: community
      provider: github
      url: https://github.com/tektoncd/catalog
      revision: main
  SCOPES: | # Update the scopes as for your requirements
    - name: agent:create
      users: [vinamra28, piyush-garg, pratap0007, puneetpunamiya, sm43, sthaha, vdemeester]
    - name: catalog:refresh
      users: [vinamra28, piyush-garg, pratap0007, puneetpunamiya, sm43, sthaha, vdemeester]
    - name: config:refresh
      users: [vinamra28, piyush-garg, pratap0007, puneetpunamiya, sm43, sthaha, vdemeester]
  DEFAULT: | # This is default scopes required to rate a resource.
    scopes:
      - rating:read
      - rating:write

All users have default scopes rating:read and rating:write. Once user get login to Hub, they get the appropriate scopes which allows them to rate the resource

Schedule Catalog Refresh On an Interval

Default Catalog refresh interval time is 30 minutes. After every 30 minutes catalog resources in the hub db would be refreshed. If you want to change catalog refresh interval, update the CATALOG_REFRESH_INTERVAL in tekton-hub-api configmap with any one of these supported time units As(A seconds), Bm(B minutes) Ch(C hours), Dd(D days) and Ew(E weeks).
If you want to skip automate catalog refresh, keep CATALOG_REFRESH_INTERVAL as empty or 0. Negative value is also not supported.

apiVersion: v1
kind: ConfigMap
metadata:
  name: tekton-hub-api
  labels:
    app: tekton-hub-api
data:
  CATALOG_REFRESH_INTERVAL: 30m ## change the catalog refresh interval here
  ...

WARN : Make sure you have updated Hub config before starting the api server

Create SSH secrets (Optional)

In order to clone private repositories or repositories from private git instances, we need to mount the ssh credentials so that cloning can happen. To make this happen please follow the below steps:

Create the ssh keys if not already present in ~/.ssh dir.

Create kubernetes secret which will contain the ssh keys using the following command

kubectl create secret generic tekton-hub-api-ssh-crds --from-file=id_rsa="/path/to/id_rsa" --from-file=id_rsa.pub="/path/to/id_rsa.pub" --from-file=known_hosts="/path/to/known_hosts"

Please make sure that secrets are created with the name tekton-hub-api-ssh-crds

Update API deployment (Optional)

By default the catalog is cloned in $HOME/catalog. If in case you want to change the clone path, edit the 02-api/22-api-deployment.yaml with the following

spec:
  ...
    spec:
      ...
      containers:
        - ...
          volumeMounts:
          - name: catalog-source
            mountPath: "/home/hub/catalog" <---- change here with path/where/catalog/to/be/cloned eg /tmp/catalog
          - name: tekton-hub-config
            mountPath: '/tmp/config' <---- path where config map to be mounted
          ...
          env:
          - name: CLONE_BASE_PATH
            value: <path/where/catalog/to/be/cloned> eg /tmp/catalog

Note: Mount path of api ConfigMap is fixed and shouldn't be changed.

Update API Image

Edit the 02-api/22-api-deployment.yaml and replace the image with the one created previously and executed below command

kubectl apply -f 02-api/ -n tekton-hub

This will create the pvc, deployment, secret, configmap and a NodePort service to expose the API server.

Verify if api route is accessible

For OpenShift:-

curl -k -X GET $(oc get -n tekton-hub routes api --template='https://{{ .spec.host }}/v1/categories')

Deploy UI

Setup Route or Ingress

After the deployment is done successfully, we need to expose the URL to access the UI.

If deploying on OpenShift:-

kubectl apply -f 04-openshift/41-ui-route.yaml -n tekton-hub

If deploying on Kubernetes:-

Create the secret containing tls cert and tls key

kubectl create secret tls ui-hub-tekton-dev-tls --cert=path/to/cert/file --key=path/to/key/file -n tekton-hub

Apply the Ingress

kubectl apply -f 04-kubernetes/41-ui-ingress.yaml -n tekton-hub

Update UI ConfigMap

Edit config/03-ui/30-ui-configmap.yaml and by setting the API URL, auth URL and Redirect URL.

You can get the API URL using below command (OpenShift)

  kubectl get -n tekton-hub routes tekton-hub-api --template='https://{{ .spec.host }}'

---
apiVersion: v1
kind: ConfigMap
metadata:
  name: tekton-hub-ui
data:
  API_URL: API URL   <<< Update this by API url
  API_VERSION: API VERSION  <<< Update this by API version (For e.g-"v1")
  AUTH_BASE_URL: AUTH URL << Update this by auth url
  REDIRECT_URI: UI URL << Update this by ui url
  CUSTOM_LOGO_MEDIA_TYPE: << Update this by your logo media type
  CUSTOM_LOGO_BASE64_DATA: << Update this by Base64 encode DATA of Custom Logo

Add Custom Logo

Create the Base64 encode value of Custom logo and update the CUSTOM_LOGO_BASE64_DATA value in the config/03-ui/30-ui-configmap.yaml and also update media type of your logo in CUSTOM_LOGO_MEDIA_TYPE, default media type is image/png .

Update UI Image

Edit the 03-ui/31-ui-deployment.yaml by replacing the image with the one created previously and executed below command

kubectl apply -f 03-ui/ -n tekton-hub

This will create the deployment, configmap and a NodePort service to expose the UI.

Ensure pods are up and running

kubectl get pods -n tekton-hub

eg. wait till status of UI pod is running

NAME                   READY   STATUS      RESTARTS   AGE
tekton-hub-api-6dfc6f97d9-vk66r   1/1     Running     3          21m
tekton-hub-db-9bd4cdf99-zsq89     1/1     Running     0          21m
tekton-hub-db-migration-26ngs     0/1     Completed   0          21m
tekton-hub-ui-55fc66cc6b-69dsp    1/1     Running     0          58s

Verify if ui route is accessible

For OpenShift:-

    kubectl get routes -n tekton-hub tekton-hub-ui --template='https://{{ .spec.host }} '

Open the URL in a browser.

Note: On UI, you should be able to see the resources on homepage and Categories on the left side.

Update your GitHub OAuth created with the UI route in place of Homepage URL and Authorization callback URL.

Update Catalog and Catalog Refresh

If you have added and modified catalog in the config file , then you will have to folllow following steps to see resources from catalog on Hub UI

Login through the Hub UI. Click on Login on right corner and then Sign in with GitHub/Github Enterprise.
Copy the Hub Token by clicking on the user profile which is at the right corner on the Home Page

Refresh the config file and you need to have config:refresh scope.

To Refresh config file

  curl -X POST -H "Authorization: <access-token>" \
  --header "Content-Type: application/json" \
  --data '{"force": true} \
  <api-route>/system/config/refresh

Call the Catalog Refresh API:
- To refresh a catalog with name
```
curl -X POST -H "Authorization: <jwt-token>" \
    <api-url>/catalog/<catalogName>/refresh
```
  Replace <access-token> with your Hub token copied from UI and replace <api-url> with your api pod url.
  
  This will give an output as below
```
[{"id":1,"catalogName":"tekton","status":"queued"}]
```
- To refresh all catalogs
```
curl -X POST -H "Authorization: <jwt-token>" \
    <api-url>/catalog/refresh
```
  Replace <access-token> with your Hub token copied from UI and replace <api-url> with your api pod url.
  
  This will give an output as below
```
[{"id":1,"catalogName":"tekton","status":"queued"}]
```
Refresh your UI, you will be able to see resources.

Setup Catalog Refresh CronJob (Optional)

You can setup a cronjob which will refresh your db after an interval if there are any changes in your catalog.

NOTE: The catalog refresh will add new resources or update existing resource if it is updated in catalog but it doesn't delete a resource in db even if it is deleted in catalog.

You will need a JWT token with catalog refresh scope which can be used.
A User token is short lived so you can create an agent token which will not expire.
To create an agent, you can use the /system/user/agent. You will need agent:create scope in your JWT which has to be added in the Api config map.
```
curl -X PUT --header "Content-Type: application/json" \
    -H "Authorization: <access-token>" \
    --data '{"name":"catalog-refresh-agent","scopes": ["catalog:refresh"]}' \
    <api-route>/system/user/agent
```
Replace <access-token> with your JWT, you can create an agent with any required scopes. In this case we need the agent with catalog:refresh. You can give any name to agent.

This will gives an output as:
```
{"token":"agent jwt token"}
```
A token will be returned with the requested scopes. You can check using the jwt decoder. Use this token for the catalog refresh cron job.
Edit 05-catalog-refresh-cj/50-catalog-refresh-secret.yaml and Set the Hub Token
```
apiVersion: v1
kind: Secret
metadata:
  name: catalog-refresh
type: Opaque
stringData:
  HUB_TOKEN: hub token
```
Set the token created in last step as Hub token.
Apply the YAMLs
```
kubectl apply -f 05-catalog-refresh-cj/ -n tekton-hub
```
The cron job is configured to run every 30 min, you can change the interval in 05-catalog-refresh-cj/51-catalog-refresh-cronjob.yaml.

Adding New Users in Config

By default when any user login for the first time, they will have only default scope even if they are added in api config map. To get additinal scopes, make sure the user has logged in once.

Now, we need to refresh the config. To do that do a POST API call.

curl -X POST -H "Authorization: <access-token>" \
    --header "Content-Type: application/json" \
    --data '{"force": true}' \
  <api-route>/system/config/refresh

Replace <access-token> with your JWT token. you must have config-refresh scope to call this API.

Deploying on Disconnected Cluster

To deploy on a disconnected cluster you need to fork the tektoncd/catalog or you can use your own catalog but it must follow the Catalog TEP.

Fork the tektoncd/hub and update the hub api config map file for your catalog details.

Update the catalog details with your catalog

catalogs:
  - name: tekton
    org: tektoncd
    type: community
    url: https://github.com/tektoncd/catalog
    revision: main

Mirror the Hub Deployment Images to the registry disconnected cluster has access to. You can use release images or build your own. For hub release images, you can use latest release tag as image tag. for eg. quay.io/tekton-hub/ui:v1.3.0. You can check the latest release [here][hub-releases].
List of images to be mirrored
- UI - quay.io/tekton-hub/ui
- API - quay.io/tekton-hub/api
- DB-Migration - quay.io/tekton-hub/db-migration
- DB - postgres:13@sha256:260a98d976574b439712c35914fdcb840755233f79f3e27ea632543f78b7a21e
You can mirror the images using oc as below
```
oc image mirror quay.io/tekton-hub/ui:v1.3.0 your.registry/project/ui:v1.3.0
```
Now, follow the steps from Deploy Database Section above. Make sure you update the images in deployment files before applying.

Troubleshooting

UI is not showing resources but catalog refresh is successful

If you have insecure connection the UI might not be able to hit the API.
So, hit the API URL in your browser https://<api-url> once you get the response as status: ok, refresh you UI.

UI is not showing resources but catalog refresh is successful (Console Error)

Verify catalog refresh is successful by hitting the /v1/resources API using the API URL.
If is returning resources than check for console errors on UI.
Right click -> Inspect or (Ctrl + Shift + I) (For Chrome). Then click on console tab.
If there is cors error then check the UI config map and the URL you have added.
The URL should be for example https://api.hub.tekton.dev and not https://api.hub.tekton.dev/ there shouldn't be / at the end.

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

DEPLOYMENT.md

DEPLOYMENT.md

Deployment

Prerequisites

Build Images

UI Image

API & DB Migration Image

NOTE: If You have deployed Hub v1.6.0 and want to upgrade to Hub v1.7.0 then please follow this doc for next steps. As we have renamed config metadata name and labels in the configs of db, api and ui.

NOTE: In case of using Github or Gitlab Enterprise make sure you deploy Hub in the same network where the Enterprise server is running. Example if your Github/Gitlab Enterprise is running behind VPN then your Kubernetes Cluster should also be behind VPN.

Deploy Database

Run DB migration

Deploy API Service

Setup Route or Ingress

Create Git Oauth Applications

Update API Secret

Update API ConfigMap

Schedule Catalog Refresh On an Interval

Create SSH secrets (Optional)

Update API deployment (Optional)

Update API Image

Verify if api route is accessible

Deploy UI

Setup Route or Ingress

Update UI ConfigMap

Add Custom Logo

Update UI Image

Verify if ui route is accessible

Update Catalog and Catalog Refresh

Setup Catalog Refresh CronJob (Optional)

Adding New Users in Config

Deploying on Disconnected Cluster

Troubleshooting

UI is not showing resources but catalog refresh is successful

UI is not showing resources but catalog refresh is successful (Console Error)

Files

DEPLOYMENT.md

Latest commit

History

DEPLOYMENT.md

File metadata and controls

Deployment

Prerequisites

Build Images

UI Image

API & DB Migration Image

NOTE: If You have deployed Hub v1.6.0 and want to upgrade to Hub v1.7.0 then please follow this doc for next steps. As we have renamed config metadata name and labels in the configs of db, api and ui.

NOTE: In case of using Github or Gitlab Enterprise make sure you deploy Hub in the same network where the Enterprise server is running. Example if your Github/Gitlab Enterprise is running behind VPN then your Kubernetes Cluster should also be behind VPN.

Deploy Database

Run DB migration

Deploy API Service

Setup Route or Ingress

Create Git Oauth Applications

Update API Secret

Update API ConfigMap

Schedule Catalog Refresh On an Interval

Create SSH secrets (Optional)

Update API deployment (Optional)

Update API Image

Verify if api route is accessible

Deploy UI

Setup Route or Ingress

Update UI ConfigMap

Add Custom Logo

Update UI Image

Verify if ui route is accessible

Update Catalog and Catalog Refresh

Setup Catalog Refresh CronJob (Optional)

Adding New Users in Config

Deploying on Disconnected Cluster

Troubleshooting

UI is not showing resources but catalog refresh is successful

UI is not showing resources but catalog refresh is successful (Console Error)