skydive-project/skydive-operator

Skydive Operator

Skydive is an open source real-time network topology and protocols analyzer found here:

• http://github.com/skydive-project/skydive

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

Pre Requisites

• 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).

Installing the Operator

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.yaml

The 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.yaml

The 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.yaml

Note: 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.

Uninstalling

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.yaml

or

$ kubectl delete -f https://github.com/skydive-project/skydive-operator/blob/master/deploy/crds/charts.helm.k8s.io_v1alpha1_netflowcollector_cr.yaml

The 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.yaml

and 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.yaml

Configuration

The 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

Documentation

Skydive documentation can be found here:

• http://skydive.network

Contact and Support

• 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

skydive-operator's Issues

UI continuously renders the topology(connecting/disconnecting) on openshift platform (using skydive operator)

I have deployed skydive on Openshift platform using skydive operator. Once the instance of operator is installed I see the resources are continuously connecting/disconnecting. Any suggestions to sort out this issue?

Skydive in OperatorHub

The goal is to contribute the skydive operator to the operator hub (https://operatorhub.io/contribute)

Initially we can use the operator-sdk generated operator from helm (https://github.com/operator-framework/operator-sdk/blob/master/doc/proposals/helm-operator.md)

Later we can develop a golang operator using kubebuilder (https://github.com/kubernetes-sigs/kubebuilder)

And going forward we could even make Skydive act like a k8s native application by creating CRDs (Custom Resource Definitions) to represent the SkydiveAnalyzer (one per cluster) and SkydiveAgent (one per node)

Browser hangs and takes too long to render the topology on openshift

I have deployed skydive operator on openshift and I am facing below problems. Please help me in resolving the issues.

1. The ui is taking too long time to load the topology and frequently browser says 'Page unresponsive' (using Google chrome and firefox).
2. Frequently I see connected/disconnected messages on the right hand top corner.
3. The agent running on the node where analyzer is running is connected to the analyzer and rest of the agents (daemon sets are unable to connect to the analyzers and the below is found in logs.

2020-08-23T18:30:51.434Z        INFO    websocket/client.go:492 (*Client).Connect       ip-10-0-143-120: Connecting to ws://172.30.75.207:8082/ws/agent
2020-08-23T18:33:02.506Z        INFO    websocket/client.go:492 (*Client).Connect       ip-10-0-143-120: Connecting to ws://172.30.75.207:8082/ws/agent
2020-08-23T18:35:13.578Z        INFO    websocket/client.go:492 (*Client).Connect       ip-10-0-143-120: Connecting to ws://172.30.75.207:8082/ws/agent
2020-08-23T18:37:24.650Z        INFO    websocket/client.go:492 (*Client).Connect       ip-10-0-143-120: Connecting to ws://172.30.75.207:8082/ws/agent
2020-08-23T18:39:35.722Z        INFO    websocket/client.go:492 (*Client).Connect       ip-10-0-143-120: Connecting to ws://172.30.75.207:8082/ws/agent
2020-08-23T18:41:46.794Z        INFO    websocket/client.go:492 (*Client).Connect       ip-10-0-143-120: Connecting to ws://172.30.75.207:8082/ws/agent
2020-08-23T18:43:57.866Z        INFO    websocket/client.go:492 (*Client).Connect       ip-10-0-143-120: Connecting to ws://172.30.75.207:8082/ws/agent

Image should be built inside the Skydive CI

skydive-operator/deploy/operator.yaml

Line 19 in 779d239

image: quay.io/sacloud/skydive-op:v0.0.2