Version: v0.6.1

Cluster Scaling

This tasks shows you how to perform a gracefull cluster scale up and scale down.

Before you begin

About this task

The Simple NiFi example consists of a three nodes NiFi cluster. A node decommission must follow a strict procedure, described in the NiFi documentation :

  1. Disconnect the node
  2. Once disconnect completes, offload the node.
  3. Once offload completes, delete the node.
  4. Once the delete request has finished, stop/remove the NiFi service on the host.

For the moment, we have implemented it as follows in the operator :

  1. Disconnect the node
  2. Once disconnect completes, offload the node.
  3. Once offload completes, delete the pod.
  4. Once the pod deletion completes, delete the node.
  5. Once the delete request has finished, remove the node from the NifiCluster status.

In addition, we have a regular check that ensure that all nodes have been removed.

In this task, you will first perform a scale up, in adding an new node. Then, you will remove another node that the one created, and observe the decommission's steps.

Scale up : Add a new node

For this task, we will simply add a node with the same configuration than the other ones, if you want to know more about how to add a node with an other configuration let's have a look to the Node configuration documentation page.

  1. Add and run a dataflow as the example :

Scaling dataflow

  1. Add a new node to the list of NifiCluster.Spec.Nodes field, by following the Node object definition documentation:
apiVersion: nifi.orange.com/v1alpha1
kind: NifiCluster
metadata:
name: simplenifi
spec:
service:
headlessEnabled: true
zkAddress: "zookeepercluster-client.zookeeper:2181"
zkPath: "/simplenifi"
clusterImage: "apache/nifi:1.12.1"
oneNifiNodePerNode: false
nodeConfigGroups:
default_group:
isNode: true
storageConfigs:
- mountPath: "/opt/nifi/nifi-current/logs"
name: logs
pvcSpec:
accessModes:
- ReadWriteOnce
storageClassName: "standard"
resources:
requests:
storage: 10Gi
serviceAccountName: "default"
resourcesRequirements:
limits:
cpu: "2"
memory: 3Gi
requests:
cpu: "1"
memory: 1Gi
nodes:
- id: 0
nodeConfigGroup: "default_group"
- id: 1
nodeConfigGroup: "default_group"
- id: 2
nodeConfigGroup: "default_group"
# >>>> START: The new node
- id: 25
nodeConfigGroup: "default_group"
# <<<< END
propagateLabels: true
nifiClusterTaskSpec:
retryDurationMinutes: 10
listenersConfig:
internalListeners:
- type: "http"
name: "http"
containerPort: 8080
- type: "cluster"
name: "cluster"
containerPort: 6007
- type: "s2s"
name: "s2s"
containerPort: 10000
important

Note : The Node.Id field must be unique in the NifiCluster.Spec.Nodes list.

  1. Apply the new NifiCluster configuration :
kubectl -n nifi apply -f config/samples/simplenificluster.yaml
  1. You should now have the following resources into kubernetes :
kubectl get pods,configmap,pvc -l nodeId=25
NAME READY STATUS RESTARTS AGE
pod/simplenifi-25-nodem5jh4 1/1 Running 0 11m
NAME DATA AGE
configmap/simplenifi-config-25 7 11m
NAME STATUS VOLUME CAPACITY ACCESS MODES STORAGECLASS AGE
persistentvolumeclaim/simplenifi-25-storagehwn24 Bound pvc-7da86076-728e-11ea-846d-42010a8400f2 10Gi RWO standard 11m

And if you go on the NiFi UI, in the cluster administration page :

Scale up, cluster list

  1. You now have data on the new node :

Scale up, cluster distribution

Scaledown : Gracefully remove node

For this task, we will simply remove a node and look at that the decommission's steps.

  1. Remove the node from the list of NifiCluster.Spec.Nodes field :
apiVersion: nifi.orange.com/v1alpha1
kind: NifiCluster
metadata:
name: simplenifi
spec:
headlessServiceEnabled: true
zkAddresse: "zookeepercluster-client.zookeeper:2181"
zkPath: "/simplenifi"
clusterImage: "apache/nifi:1.11.3"
oneNifiNodePerNode: false
nodeConfigGroups:
default_group:
isNode: true
storageConfigs:
- mountPath: "/opt/nifi/nifi-current/logs"
name: logs
pvcSpec:
accessModes:
- ReadWriteOnce
storageClassName: "standard"
resources:
requests:
storage: 10Gi
serviceAccountName: "default"
resourcesRequirements:
limits:
cpu: "2"
memory: 3Gi
requests:
cpu: "1"
memory: 1Gi
nodes:
- id: 0
nodeConfigGroup: "default_group"
- id: 1
nodeConfigGroup: "default_group"
# >>>> START: node removed
# - id: 2
# nodeConfigGroup: "default_group"
# <<<< END
- id: 25
nodeConfigGroup: "default_group"
propagateLabels: true
nifiClusterTaskSpec:
retryDurationMinutes: 10
listenersConfig:
internalListeners:
- type: "http"
name: "http"
containerPort: 8080
- type: "cluster"
name: "cluster"
containerPort: 6007
- type: "s2s"
name: "s2s"
containerPort: 10000
  1. Apply the new NifiCluster configuration :
kubectl -n nifi apply -f config/samples/simplenificluster.yaml
  1. You can follow the node's action step status in the NifiCluster.Status description :
kubectl describe nificluster simplenifi
...
Status:
Nodes State:
...
2:
Configuration State: ConfigInSync
Graceful Action State:
Action State: GracefulDownscaleRequired
Error Message:
...
...
tip

The list of decommision's step and their corresponding value for the Nifi Cluster.Status.Node State.Graceful ActionState.ActionStep field is described into the Node State page

  1. Once the scaledown successfully performed, you should have the data offloaded on the other nodes, and the node state removed from the NifiCluster.Status.NodesState list :
warning

Keep in mind that the NifiCluster.Spec.nifiClusterTaskSpec.retryDurationMinutes should be long enough to perform the whole procedure, or you will have some rollback and retry loop.

Last updated on by Alexandre Guitton