Codefresh on-premises platform upgrade

Upgrade the Codefresh On-Premises platform

Upgrade the Codefresh on-premises platform to the latest version:

  • Prepare for the upgrade: Before the upgrade, based on the version you are upgrading to, complete the required tasks
  • Upgrade the platform
  • Complete post-upgrade configuration: If needed, also based on the version you are upgrading to, complete the required tasks

Upgrade to 1.2.0 and higher

This major release deprecates the following Codefresh managed charts:

  • Ingress
  • Rabbitmq
  • Redis

See the instructions below for each of the affected charts.

NOTES
Before the upgrade:

  • Remove any seed jobs left from previous release with: kubectl delete job --namespace ${CF_NAMESPACE} -l release=cf

  • Remove PDBs for Redis and RabbitMQ left from previous release with:
    kubectl delete pdb cf-rabbitmq --namespace ${CF_NAMESPACE}
    kubectl delete pdb cf-redis --namespace ${CF_NAMESPACE}

Update configuration for Ingress chart

From version 1.2.0 and higher, we have deprecated support for Codefresh-managed-ingress. Kubernetes community public ingress-nginx chart replaces Codefresh-managed-ingress chart. For more information on the ingress-nginx, see kubernetes/ingress-nginx.

NOTE
Parameter locations have changed as the ingress chart name was changed from ingress to ingress-nginx: NGINX controller parameters are now defined under ingress-nginx Ingress object parameters are now defined under ingress

You must update config.yaml, if you are using:

  • External ingress controllers, including ALB (Application Load Balancer)
  • Codefresh-managed ingress controller with custom values

Update configuration for external ingress controllers

For external ingress controllers, including ALB (Application Load Balancer), update the relevant sections in config.yaml to align with the new name for the ingress chart:

  • Replace ingress with ingress-nginx

v1.1.1 or lower

ingress: #disables creation of both Nginx controller deployment and Ingress objects
  enabled: false

v1.2.2 or higher

ingress-nginx: #disables creation of Nginx controller deployment
  enabled: false

ingress: #disables creation of Ingress objects (assuming you've manually created ingress resource before)
  enabled: false
  • Replace annotations that have been deprecated with ingressClassName

v1.1.1 or lower

ingress:
  annotations:
    kubernetes.io/ingress.class: my-non-codefresh-nginx

v1.2.2 or higher

ingress-nginx:
  enabled: false

ingress:
  ingressClassName: my-non-codefresh-nginx
###  `kubernetes.io/ingress.class` annotation is deprecated from kubernetes v1.22+.
#  annotations:
#    kubernetes.io/ingress.class: my-non-codefresh-nginx

Update configuration for Codefresh-managed ingress with custom values

If you were running Codefresh-managed ingress controller with custom values refer to values.yaml from the official repo. If needed, update the ingress-nginx section in config.yaml. The example below shows the default values (already provided in Codefresh chart) for ingress-nginx:

ingress-nginx:
  enabled: true
  controller:
    ## This section refers to the creation of the IngressClass resource
    ## IngressClass resources are supported since k8s >= 1.18 and required since k8s >= 1.19
    ingressClassResource:
      # -- Is this ingressClass enabled or not
      enabled: true
      # -- Is this the default ingressClass for the cluster
      default: false
      # -- Controller-value of the controller that is processing this ingressClass
      controllerValue: "k8s.io/ingress-nginx-codefresh"
      # -- Name of the ingressClass
      name: nginx-codefresh
    # -- For backwards compatibility with ingress.class annotation.
    # Algorithm is as follows, first ingressClassName is considered, if not present, controller looks for ingress.class annotation
    ingressClass: nginx-codefresh
    # -- Process IngressClass per name (additionally as per spec.controller).
    ingressClassByName: true
    # Limit the scope of the controller to a specific namespace
    scope:
      # -- Enable 'scope' or not
      enabled: true
    admissionWebhooks:
      enabled: false

NOTE
New ingress-nginx subchart creates a new cf-ingress-nginx-controller service (type: LoadBalancer) instead of old cf-ingress-controller service. So make sure to update DNS record for global.appUrl to point to a new external load balancer IP. You can get external load balancer IP with: kubectl get svc cf-ingress-nginx-controller -o jsonpath={.status.loadBalancer.ingress[0].ip

Update configuration for RabbitMQ chart

From version 1.2.2 and higher, we have deprecated support for the Codefresh-managed Rabbitmq chart. Bitnami public bitnami/rabbitmq chart has replaced the Codefresh-managed rabbitmq. For more information, see bitnami/rabbitmq.

NOTE
Configuration updates are not required when running an external RabbitMQ service.

RabbitMQ chart was replaced so as a consequence values structure might be different for some parameters. For the complete list of values, see values.yaml

existingPvc changed to existingClaim and defined under persistence

v1.1.1 or lower

rabbitmq:
  existingPvc: my-rabbitmq-pvc
  nodeSelector:
    foo: bar
  resources:
    limits:
      cpu: 2000m
      memory: 2Gi
    requests:
      cpu: 500m
      memory: 1Gi
  tolerations:
  - effect: NoSchedule
    key: <key>
    operator: Equal
    value: <value>

v1.2.2 or higher

rabbitmq:
  volumePermissions: ## Enable init container that changes the owner and group of the persistent volume from existing claim
    enabled: true
  persistence:
    existingClaim: my-rabbitmq-pvc
  nodeSelector:
    foo: bar
  resources:
    limits:
      cpu: 2000m
      memory: 2Gi
    requests:
      cpu: 500m
      memory: 1Gi
  tolerations:
  - effect: NoSchedule
    key: <key>
    operator: Equal
    value: <value>

storageClass and size defined under persistence

v1.1.1 or lower

rabbitmq:
  storageClass: my-storage-class
  storageSize: 32Gi

v1.2.2 or higher

rabbitmq:
  persistence:
    storageClass: my-storage-class
    size: 32Gi

Update configuration for Redis chart

From version 1.2.2 and higher, we have deprecated support for the Codefresh-managed Redis chart. Bitnami public bitnami/redis chart has replaced the Codefresh-managed Redis chart. For more information, see bitnami/redis.

Redis storage contains CRON and Registry typed triggers so you must migrate existing data from the old deployment to the new stateful set. This is done by backing up the existing data before upgrade, and then restoring the backed up data after upgrade.

NOTE
Configuration updates are not required:

  • When running an external Redis service.
  • If CRON and Registy triggers have not been configured.

Verify existing Redis data for CRON and Registry triggers

Check if you have CRON and Registry triggers configured in Redis.

  • Run codefresh get triggers OR Directly from the K8s cluster where Codefresh is installed.
NAMESPACE=codefresh
REDIS_PASSWORD=$(kubectl get secret --namespace $NAMESPACE cf-redis -o jsonpath="{.data.redis-password}" | base64 --decode)

kubectl exec -it deploy/cf-redis -- env REDIS_PASSWORD=$REDIS_PASSWORD bash
#once inside cf-redis pod
REDISCLI_AUTH="$REDIS_PASSWORD" redis-cli
info keyspace # list db
select 15 # select db 15
keys * #show keys
  • If there are results, continue with Back up existing Redis data.

Back up existing Redis data

Back up the existing data before the upgrade:

  • Connect to the pod, run redis-cli, export AOF data from old cf-redis-* pod:
NAMESPACE=codefresh
REDIS_PASSWORD=$(kubectl get secret --namespace $NAMESPACE cf-redis -o jsonpath="{.data.redis-password}" | base64 --decode)
REDIS_POD=$(kubectl get pods -l app=cf-redis -o custom-columns=:metadata.name --no-headers=true)
kubectl cp $REDIS_POD:/bitnami/redis/data/appendonly.aof appendonly.aof -c cf-redis

Restore backed-up Redis data

Restore the data after the upgrade:

  • Copy appendonly.aof to the new cf-redis-master-0 pod:

    kubectl cp appendonly.aof cf-redis-master-0:/data/appendonly.aof
    
  • Restart cf-redis-master-0 and cf-api pods:

    kubectl delete pod cf-redis-master-0
    
    kubectl scale deployment cf-cfapi-base --replicas=0 -n codefresh
    kubectl scale deployment cf-cfapi-base --replicas=2 -n codefresh
    

NOTE
Redis chart was replaced so as a consequence values structure might be different for some parameters.
For the complete list of values, see values.yaml.

existingPvc changed to existingClaim and defined under master.persistence

v1.1.1 or lower

redis:
  existingPvc: my-redis-pvc
  nodeSelector:
    foo: bar
  resources:
    limits:
      cpu: 1000m
      memory: 1Gi
    requests:
      cpu: 500m
      memory: 500Mi
  tolerations:
  - effect: NoSchedule
    key: <key>
    operator: Equal
    value: <value>

v1.2.2 or higher

redis:
  volumePermissions: ## Enable init container that changes the owner and group of the persistent volume from existing claim
    enabled: true
  master:
    persistence:
      existingClaim: my-redis-pvc
    nodeSelector:
      foo: bar
    resources:
      limits:
        cpu: 1000m
        memory: 1Gi
      requests:
        cpu: 500m
        memory: 500Mi
    tolerations:
    - effect: NoSchedule
      key: <key>
      operator: Equal
      value: <value>

storageClass and size defined under master.persistence

v1.1.1 or lower

redis:
  storageClass: my-storage-class
  storageSize: 32Gi

v1.2.2 or higher

redis:
  master:
    persistence:
      storageClass: my-storage-class
      size: 32Gi

NOTE
If you run the upgrade without Redis backup and restore procedure, Helm Releases Dashboard page might be empty for a few minutes after the upgrade.

Upgrade to 1.3.0 and higher

This major release deprecates the following Codefresh managed charts:

  • Consul
  • Nats

Update configuration for Consul

From version 1.3.0 and higher, we have deprecated the Codefresh-managed consul chart, in favor of Bitnami public bitnami/consul chart. For more information, see bitnami/consul.

Consul storage contains data about Windows worker nodes, so if you had any Windows nodes connected to your OnPrem installation, see the following instruction:

NOTE
Use https://<your_onprem_domain>/admin/nodes to check for any existing Windows nodes.

Back up existing consul data

Before starting the upgrade, back up existing data.

NOTE
Because cf-consul is a StatefulSet and has some immutable fields in its spec with both old and new charts having the same names, you cannot perform a direct upgrade.

Direct upgrade will most likely fail with:
helm.go:84: [debug] cannot patch "cf-consul" with kind StatefulSet: StatefulSet.apps "cf-consul" is invalid: spec: Forbidden: updates to statefulset spec for fields other than 'replicas', 'template', 'updateStrategy' and 'minReadySeconds' are forbidden

After backing up existing data, you must delete the old StatefulSet.

  1. Exec into the consul pod and create a snapshot:
    kubectl exec -it cf-consul-0 -n codefresh -- consul snapshot save backup.snap
    
  2. Copy snapshot locally:
    kubectl cp -n codefresh cf-consul-0:backup.snap backup.snap
    
  3. Delete the old cf-consul stateful set:
kubectl delete statefulset cf-consul -n codefresh

Restore backed up data

After completing the upgrade to the current version, restore the consul data that you backed up.

  1. Copy the snapshot back to the new pod:
kubectl cp -n codefresh backup.snap cf-consul-0:/tmp/backup.snap
  1. Restore the data:
    kubectl exec -it cf-consul-0 -n codefresh -- consul snapshot restore /tmp/backup.snap
    

    NOTE
    Consul chart was replaced, and values structure might be different for some parameters.
    For the complete list of values, see values.yaml

Update Nats configuration

From version 1.3.0 and higher, we have deprecated Codefresh-managed nats chart in favor of Bitnami public bitnami/nats chart. For more information, see bitnami/nats.

NOTE
Because cf-nats is a StatefulSet and has some immutable fields in its spec, both the old and new charts have the same names, preventing a direct upgrade.

Direct upgrade will most likely fail with:
helm.go:84: [debug] cannot patch "cf-nats" with kind StatefulSet: StatefulSet.apps "cf-nats" is invalid: spec: Forbidden: updates to statefulset spec for fields other than 'replicas', 'template', 'updateStrategy' and 'minReadySeconds' are forbidden

  • Delete the old cf-nats stateful set.
kubectl delete statefulset cf-nats -n codefresh

NOTE
Nats chart was replaced, and values structure might be different for some parameters.
For the complete list of values, see values.yaml.

Upgrade to 1.3.1 and higher

Chart v1.3.1 fixes duplicated env vars CLUSTER_PROVIDERS_URI and CLUSTER_PROVIDERS_PORT in cf-api deployment.

W1010 03:03:55.553842     280 warnings.go:70] spec.template.spec.containers[0].env[94].name: duplicate name "CLUSTER_PROVIDERS_URI"
W1010 03:03:55.553858     280 warnings.go:70] spec.template.spec.containers[0].env[95].name: duplicate name "CLUSTER_PROVIDERS_PORT"

NOTE
Due to Helm issue Removal of duplicate array entry removes completely from Kubernetes, you shoud run kcfi deploy or helm upgrade two times consecutively.

With chart v1.3.1 insecure registy property has been moved under builder section:

builder:
  insecureRegistries:
  - "myregistrydomain.com:5000"

Upgrade to 1.4.0 and higher

Affected values:

  • HorizontalPodAutoscaler is renamed to hpa

NOTE
Update kcfi tool to the latest 0.5.18 version.

This major release deprecates the following Codefresh managed charts and replaces them with Bitnami charts:

NOTE
Read instructions before the upgrade.

IMPORTANT
There is a known issue with releases 1.4.0-1.4.6, where ABAC feature for git contexts doesn’t work as expected and has to be disabled.
Two options to mitigate the issue is to install patch release 1.4.7 or higher, where the issue is fixed or to apply a manual workaround:

  • To do that, sign in with a platform admin user and proceed to
    https://<YOUR_ONPREM_URL>/admin/features-management?filter=search:abacGitContext page
  • On the page enable System Features switch (confirm in the pop up window)
  • After that, abacGitContext feature appears on the screen, toggle it’s switch ON and then back to OFF state.

Update configuration for Postgresql chart

From version 1.4.0 and higher, we have deprecated support for the Codefresh-managed Postgresql chart. It has been replaced with Bitnami public chart bitnami/postgresql.
For more information, see bitnami/postgresql.

NOTE
If in config.yaml, postgresql.enabled=false indicating that you are running and external Postgresql service, you can ignore the configuration instructions.

IMPORTANT
Run the upgrade with global.seedJobs=true flag:

global:
  seedJobs: true

Manual backup and restore

Before the upgrade:

  1. Obtain the PostgresSQL administrator password:
    NAMESPACE=codefresh
    export PGPASSWORD=$(kubectl get secret --namespace $NAMESPACE cf-postgresql -o jsonpath="{.data.postgres-password}" | base64 --decode)
    
  2. Forward the PostgreSQL service port and place the process in the background:
    kubectl port-forward --namespace $NAMESPACE svc/cf-postgresql 5432:5432 &
    
  3. Create a directory for the backup files and make it the current working directory:
    mkdir psqlbackup
    chmod o+w psqlbackup
    cd psqlbackup
    
  4. Back up the contents of audit database to the current directory using the pg_dump tool.
    If this tool is not installed on your system, use Bitnami’s PostgreSQL Docker image to perform the backup, as shown below:
    docker run --rm --name postgresql-backup -e PGPASSWORD=$PGPASSWORD -v $(pwd):/app --net="host" bitnami/postgresql:13 pg_dump -Fc --dbname audit -h host.docker.internal -p 5432 -f /app/audit.dump
    

NOTE
The above command is true for Windows and macOS, for Linux users host.docker.internal has to be replaced with 127.0.0.1.

Here:

  • The –net parameter lets the Docker container use the host’s network stack and thereby gain access to the forwarded port.
  • The pg_dump command connects to the PostgreSQL service and creates backup files in the /app directory, which is mapped to the current directory (psqlbackup/) on the Docker host with the -v parameter.
  • The –rm parameter deletes the container after the pg_dump command completes execution.

After the upgrade:

Create an environment variable with the password for the new stateful set:

export PGPASSWORD=$(kubectl get secret --namespace $NAMESPACE cf-postgresql -o jsonpath="{.data.postgres-password}" | base64 --decode)

Forward the PostgreSQL service port for the new stateful set and place the process in the background:

kubectl port-forward --namespace $NAMESPACE svc/cf-postgresql 5432:5432

Restore the contents of the backup into the new release using the pg_restore tool. If this tool is not available on your system, mount the directory containing the backup files as a volume in Bitnami’s PostgreSQL Docker container and use the pg_restore client tool in the container image to import the backup into the new cluster, as shown below:

cd psqlbackup
docker run --rm --name postgresql-backup -e PGPASSWORD=$PGPASSWORD -v $(pwd):/app --net="host" bitnami/postgresql:13 pg_restore -Fc --create --dbname postgres -h host.docker.internal -p 5432 /app/audit.dump

NOTE
The above command is true for Windows and macOS, for Linux users host.docker.internal must be replaced with 127.0.0.1.

Backup and restore via Helm hooks

You can also run Postgresql database migration with pre-upgrade and post-upgrade helm hooks.

NOTE
It’s strongly recommended to create a MANUAL backup prior to the upgrade!

To enable backup and restore via Helm hooks:

  • Set postgresql.migration.enabled=true (specify additional parameters if necessary). It will create a k8s Job with PVC, and run pg_dump and pg_restore during the upgrade.
postgresql:
  migration:
    ## enable the migration Job (pre-upgrade and post-upgrade hooks)
    enabled: true
    ## specify Storage Class for PVC which will store the backup (by default will use the default SC on your k8s cluster)
    # storageClass: standard
    ## specify PVC size (default size is 8Gi, make sure to adjust the size to the current config `postgresql.persistence.size` value)
    # pvcSize: 8Gi
    ## set Job to use an existing PVC
    existingClaim: ""
    ## set nodeSelector/tolerations/resources for the Job
    nodeSelector: {}
    tolerations: []
    resources: {}

Update configuration for MongoDB chart

From version 1.4.0 and higher, we have deprecated support for the Codefresh-managed MongoDB chart. It has been replaced with Bitnami public chart bitnami/mongodb.
For more information, see bitnami/mongodb.

NOTE
If in config.yaml, mongo.enabled=false, indicating that you are running an external MongoDB service, you can ignore the configuration updates.

IMPORTANT
Run the upgrade with global.seedJobs=true flag:

global:
  seedJobs: true

Manual backup and restore

Before the upgrade:

  1. Obtain the MongoDB administrator password:
    NAMESPACE=codefresh
    export MONGODB_ROOT_PASSWORD=$(kubectl get secret --namespace $NAMESPACE cf-mongodb -o jsonpath="{.data.mongodb-root-password}" | base64 --decode)
    
  2. Forward the MongoDB service port and place the process in the background:
    kubectl port-forward --namespace $NAMESPACE svc/mongodb 27017:27017 &
    
  3. Create a directory for the backup files and make it the current working directory:
    mkdir mongobackup
    chmod o+w mongobackup
    cd mongobackup
    
  4. Back up the contents of all the databases to the current directory using the monogodump tool.
    If this tool is not installed on your system, use Bitnami’s MongoDB Docker image to perform the backup, as shown below:
    docker run --rm --name mongodb-backup -v $(pwd):/app --net="host" bitnami/mongodb:4.2 mongodump --host="host.docker.internal:27017" -u root -p $MONGODB_ROOT_PASSWORD -o /app
    

NOTE
The above command is true for Windows and macOS, for Linux users host.docker.internal has to be replaced with 127.0.0.1.

Here:

  • The –net parameter lets the Docker container use the host’s network stack and thereby gain access to the forwarded port.
  • The mongodump command connects to the MongoDB service and creates backup files in the /app directory, which is mapped to the current directory (mongobackup/) on the Docker host with the -v parameter.
  • The –rm parameter deletes the container after the mongodump command completes execution.

After the upgrade:

  1. Create an environment variable with the password for the new stateful set:
    export MONGODB_ROOT_PASSWORD=$(kubectl get secret --namespace $NAMESPACE cf-mongodb -o jsonpath="{.data.mongodb-root-password}" | base64 --decode)
    
  2. Forward the MongoDB service port for the new stateful set and place the process in the background: (Note! Default service address was changed)
    kubectl port-forward --namespace $NAMESPACE svc/cf-mongodb 27017:27017 &
    
  3. Restore the contents of the backup into the new release using the mongorestore tool.
    If this tool is not available on your system, mount the directory containing the backup files as a volume in Bitnami’s MongoDB Docker container and use the mongorestore client tool in the container image to import the backup into the new cluster, as shown below:
    cd mondgobackup
    docker run --rm --name mongodb-backup -v $(pwd):/app --net="host" bitnami/mongodb:4.2 mongorestore --host="host.docker.internal:27017" -u root -p $MONGODB_ROOT_PASSWORD /app
    

NOTE
The above command is true for Windows and macOS, for Linux users host.docker.internal has to be replaced with 127.0.0.1.

  1. Stop the service port forwarding by terminating the background process.
  2. Connect to the new stateful set and confirm that your data has been successfully restored:
    kubectl run --namespace $NAMESPACE mongodb-new-client --rm --tty -i --restart='Never' --image docker.io/bitnami/mongodb:4.2 --command -- mongo codefresh --host cf-mongodb --authenticationDatabase admin -u root -p $MONGODB_ROOT_PASSWORD --eval "db.accounts.find()"
    

Backup and restore via Helm hooks

You can also run MongoDB database migration with pre-upgrade and post-upgrade helm hooks.

NOTE
It’s strongly recommended to create a MANUAL backup prior to the upgrade!

To enable backup and restore via Helm hooks:

  • Set mongodb.migration.enabled=true (specify additional parameters if necessary).
    It will create a k8s Job with PVC and run mongodump and mongorestore during the upgrade.
mongodb:
  migration:
    ## enable the migration Job (pre-upgrade and post-upgrade hooks)
    enabled: true
    ## specify Storage Class for PVC which will store the backup (by default will use the default SC on your k8s cluster)
    # storageClass: standard
    ## specify PVC size (default size is 8Gi, make sure to adjust the size to the current config `mongodb.persistence.size` value)
    # pvcSize: 8Gi
    ## set Job to use an existing PVC
    existingClaim: ""
    ## set nodeSelector/tolerations/resources for the Job
    nodeSelector: {}
    tolerations: []
    resources: {}

Upgrade to 2.0.0

Version 2.0.0 incorporates a major version and chart change, including breaking changes that need you to take actions manually for compatibility.

WARNING
The kcfiinstaller has been deprecated from Version 2.0.0 and higher.

Helm is the recommended way to install Codefresh On-Premises. The kcfi config.yaml is NOT compatible with Helm-based installation. To reuse the same config.yaml for the Helm chart, you need to remove deprecated sections, and update configuration for other sections.

Follow the instructions in Upgrading to 2.0.0.

Upgrade to 2.2.0

Version 2.2.0 incorporates changes.

Follow the instructions in Upgrading to 2.2.0.

Codefresh on-premises setup
On-premises feature management