Skydive is an open source real-time network topology and protocols analyzer found here:
This operator includes two default installation options of Skydive:
- A full Skydive version
- Skydive as netflow collector
Each option defines the default parameters required, and can be used as is for testing. For production it is adviced to customize the defaults.
This document shows how to install the skydive-operator, and deploy the Skydive different options
-
Kubernetes version 1.10.0 and higher
-
Persistent volume is needed only if you want to "look back in time" with skydive (that is, if you are interested in the monitoring history); if you don't , then it is not required (an elastic search container will not be created). You can create a persistent volume via the IBM Cloud Private interface or through a yaml file. For example:
apiVersion: v1
kind: PersistentVolume
metadata:
name: <persistent volume name>
spec:
capacity:
storage: 10Gi
accessModes:
- ReadWriteOnce
hostPath:
path: <PATH>
- S3 compatible Object Storage available either locally on the same cluster, or at some other location. It is required only if you want to store the netflows collected by skydive for some future analysis; if you don't , then it is not required (an exporter container will not be created).
To install the operator:
First deploy the custom resource definitions:
$ kubectl create -f https://raw.githubusercontent.com/skydive-project/skydive-operator/master/deploy/crds/charts.helm.k8s.io_skydives_crd.yaml
$ kubectl create -f https://raw.githubusercontent.com/skydive-project/skydive-operator/master/deploy/crds/charts.helm.k8s.io_netflowcollectors_crd.yamlThe charts.helm.k8s.io_skydives_crd.yaml defines a resource type skydive and charts.helm.k8s.io_netflowcollectors_crd.yaml defines skydive as a netflow collector
then install the operator:
$ kubectl create -f https://raw.githubusercontent.com/skydive-project/skydive-operator/master/deploy/service_account.yaml
$ kubectl create -f https://raw.githubusercontent.com/skydive-project/skydive-operator/master/deploy/role.yaml
$ kubectl create -f https://raw.githubusercontent.com/skydive-project/skydive-operator/master/deploy/role_binding.yaml
$ kubectl create -f https://raw.githubusercontent.com/skydive-project/skydive-operator/master/deploy/operator.yamlThe commands deploy skydive-operator on the Kubernetes cluster. The operator now will monitor the defined custom resources (netflowcollectors and skydives). At this point there are no such actual resources present in the cluster, only definitions of the 2 possible types.
Next an actuall resource can be created - either one of the two predefined possible types:
- To create a general Skydive resource run:
kubectl create -f https://raw.githubusercontent.com/skydive-project/skydive-operator/master/deploy/crds/charts.helm.k8s.io_v1alpha1_skydive_cr.yaml- To create Skydive as netflow collector resource run:
kubectl create -f https://raw.githubusercontent.com/skydive-project/skydive-operator/master/deploy/crds/charts.helm.k8s.io_v1alpha1_netflowcollector_cr.yamlNote: By running on of the above commands, you instruct the skydive-operator to deploy the selected option of skydive with default configuration parameters. To customize the configuration parameters, copy the rellevant cr yaml file to your local environment, modify the values (see the Configuration section below) and run the create command on the modified file.
To remove the custom resource:
$ kubectl delete -f https://github.com/skydive-project/skydive-operator/blob/master/deploy/crds/charts.helm.k8s.io_v1alpha1_skydive_cr.yamlor
$ kubectl delete -f https://github.com/skydive-project/skydive-operator/blob/master/deploy/crds/charts.helm.k8s.io_v1alpha1_netflowcollector_cr.yamlThe command causes the skydive-operator to remove all the Kubernetes components associated with skydive custom resource/ netflow collector custom resource
then (optionally) you can remove the operator:
$. kubectl delete -f https://raw.githubusercontent.com/skydive-project/skydive-operator/master/deploy/operator.yamland the custom resource definitions and rbac definitions:
$ kubectl delete -f https://raw.githubusercontent.com/skydive-project/skydive-operator/master/deploy/crds/charts.helm.k8s.io_skydives_crd.yaml
$ kubectl delete -f https://raw.githubusercontent.com/skydive-project/skydive-operator/master/deploy/crds/charts.helm.k8s.io_netflowcollectors_crd.yaml
$ kubectl delete -f https://raw.githubusercontent.com/skydive-project/skydive-operator/master/deploy/role_binding.yaml
$ kubectl delete -f https://raw.githubusercontent.com/skydive-project/skydive-operator/master/deploy/role.yaml
$ kubectl delete -f https://raw.githubusercontent.com/skydive-project/skydive-operator/master/deploy/service_account.yamlThe netflowcollector custom resource can be customized using the following configuration parameters:
| Parameter | Description | Default |
|---|---|---|
exporter.write.s3.endpoint |
Endpoint of the Object Storage to be used | http://127.0.0.1:9000 |
exporter.write.s3.installLocalMinio |
Install default Minio Object Storage locally | true |
exporter.write.s3.region |
Object Store region | default |
exporter.write.s3.use_api_key |
Use an api key for Object Store autentication | false |
exporter.write.s3.api_key |
api key for Object Store autentication | Empty |
exporter.write.s3.access_key |
access key for Object Store autentication | admin |
exporter.write.s3.secret_key |
secret key for Object Store autentication | admin1234 |
exporter.store.bucket |
bucket name to be used in Object Store | default |
exporter.store.objectPrefix |
prefix of stroed objects | default |
exporter.classify.cluster_net_masks |
masks of inner cluster ips (space sepparated), used for classification of netflows to ingress/egress/internal/other | 10.0.0.0/8 172.16.0.0/12 192.168.0.0/16 |
For testing purpuses the default values can be kept, for production environment
- exporter.write.s3.installLocalMinio should be set to false
- endpoint should be set to an S3 compatible Object Storage
- the api_key or access_key and secret_key sould be set (set use_api_key to true, if you provide the api_key)
- set exporter.store.bucket
- exporter.store.objectPrefix can be set to simulate a 'directory' location of the collected data inside the bucket.
The Skydive custom resource can be castomized using the following configuration parameters (and generally using any of the helm configuration parameters definedhttps://github.com/skydive-project/skydive-helm.):
| Parameter | Description | Default |
|---|---|---|
image.repository |
Skydive image repository | skydive/skydive |
image.tag |
Image tag | 0.24.0 |
image.secretName |
Image secret for private repository | Empty |
image.imagePullPolicy |
Image pull policy | IfNotPresent |
resources |
CPU/Memory resource requests/limits | Memory: 8192Mi, CPU: 2000m |
service.name |
service name | skydive |
service.type |
k8s service type (e.g. NodePort, LoadBalancer) | NodePort |
service.port |
TCP port | 8082 |
etcd.port |
TCP port | 12379 |
analyzer.topology.fabric |
Fabric connecting k8s nodes | TOR1->*[Type=host]/eth0 |
env |
Extended environment variables | Empty |
storage.elasticsearch.host |
ElasticSearch host | 127.0.0.1 |
storage.elasticsearch.port |
ElasticSearch port | 9200 |
storage.flows.indicesToKeep |
Number of flow indices to keep in storage | 10 |
storage.flows.indexEntriesLimit |
Number of flow records to keep per index | 10000 |
storage.topology.indicesToKeep |
Number of topology indices to keep in storage | 10 |
storage.topology.indexEntriesLimit |
Number of topology records to keep per index | 10000 |
persistence.enabled |
Use a PVC to persist data | false |
persistence.useDynamicProvisioning |
Specify a storageclass or leave empty | false |
dataVolume.name |
Name of the PVC to be created | datavolume |
dataVolume.existingClaimName |
Provide an existing PersistentVolumeClaim | nil |
dataVolume.storageClassName |
Storage class of backing PVC | nil |
dataVolume.size |
Size of data volume | 10Gi |
Skydive documentation can be found here:
- Mailing list: https://www.redhat.com/mailman/listinfo/skydive-dev
- Issues: https://github.com/skydive-project/skydive/issues https://github.com/skydive-project/skydive-operator/issues
- Slack
Invite : https://slack.skydive.network Workspace : https://skydive-project.slack.com