jupyterhub-istio-proxy
is a scalable solution for Jupyterhub's high network traffic demands. It implements the jupyterhub proxy api to configure istio based on requests from hub.
The following requests are supported:
GET /api/routes
: Gets all routes that have been configured on istioPOST /api/routes/<path>
: Add the route to istioDELETE /api/routes/<path>
: Remove the route from istioSince the proxy is stateless, it can be scaled horizontally. Multiple replicas can be used to ensure uptime during deployments and handle pod failure.
In order to use the jupyterhub-istio-proxy
the following prerequisites need to be met.
jupyterhub-istio-proxy
should have ability to list, get, create and delete istio virtual services in the namespace where the deployment is done. Refer Kubernetes RBAC for details.Unlike the default configurable-http-proxy
that ships with Jupyterhub, the traffic is not routed through the proxy itself. The proxy configures istio to handle all traffic to the notebook servers as well as Jupyterhub. As a result, the proxy-public
service is not needed when using jupyterhub-istio-proxy
. For more information see https://medium.com/@harsimran.maan/running-jupyterhub-with-istio-service-mesh-on-kubernetes-a-troubleshooting-journey-707039f36a7b
The proxy can be deployed to a Kubernetes namespace running Jupyterhub by applying the following config: Change SUB_DOMAIN_HOST to a value to a hostname where jupyterhub is hosted. The ISTIO_GATEWAY value should be set to the gateway which handles traffic for jupyterhub. If your cluster has a non-default domain, you can specify that with CLUSTER_DOMAIN
apiVersion: apps/v1
kind: Deployment
metadata:
labels:
app: jupyterhub
component: proxy
name: proxy
spec:
replicas: 3
selector:
matchLabels:
app: jupyterhub
component: proxy
release: RELEASE-NAME
strategy:
type: RollingUpdate
template:
metadata:
labels:
app: jupyterhub
component: proxy
release: RELEASE-NAME
spec:
affinity:
podAntiAffinity:
preferredDuringSchedulingIgnoredDuringExecution:
- podAffinityTerm:
labelSelector:
matchExpressions:
- key: name
operator: In
values:
- proxy
topologyKey: kubernetes.io/hostname
weight: 100
containers:
- command:
- /proxy/jupyterhub-istio-proxy
env:
- name: CONFIGPROXY_AUTH_TOKEN
valueFrom:
secretKeyRef:
key: proxy.token
name: hub-secret
- name: ISTIO_GATEWAY
value: jupyterhub-gateway
- name: K8S_NAMESPACE
valueFrom:
fieldRef:
fieldPath: metadata.namespace
- name: SUB_DOMAIN_HOST
value: '*'
- name: VIRTUAL_SERVICE_PREFIX
value: jupyterhub
- name: WAIT_FOR_WARMUP
value: "true"
- name: CLUSTER_DOMAIN
value: "cluster.local"
image: splunk/jupyterhub-istio-proxy:0.3.0
imagePullPolicy: IfNotPresent
name: proxy
ports:
- containerPort: 8000
name: proxy-api
protocol: TCP
resources:
limits:
cpu: "1"
memory: 256M
requests:
cpu: 100m
memory: 256M
securityContext:
allowPrivilegeEscalation: false
securityContext:
runAsNonRoot: true
terminationGracePeriodSeconds: 60
---
apiVersion: v1
kind: Service
metadata:
name: proxy-api
spec:
ports:
- name: http-proxy-api
port: 8001
protocol: TCP
targetPort: 8000
selector:
component: proxy
type: ClusterIP
---
Jupyterhub user pod creation flow when using jupyterhub-istio-proxy
.
The project requires Go 1.17.3+. Run the tests
make test
https://github.com/golang/mock is used for creating mocks for testing.
mockgen --source=proxy/istio.go -destination=proxy/istio_mock_test.go -write_package_comment -package=proxy