NOTE: This is the knative branch of Dispatch. Full Dispatch functionality is still a ways off. The code here represents a work in progress. For information about Dispatch Solo, the version of Dispatch distributed as an OVA, see the
solo
branch or documentation.
Dispatch is a framework for deploying and managing serverless style applications. The intent is a framework which enables developers to build applications which are defined by functions which handle business logic and services which provide all other functionality:
- State (Databases)
- Messaging/Eventing (Queues)
- Ingress (Api-Gateways)
- Etc.
Our goal is to provide a substrate which can be built upon and extended to serve as a framework for serverless applications. Additionally, the framework must provide tools and features which aid the developer in building, debugging and maintaining their serverless application.
Checkout the detailed documentation including a quickstart guide.
NOTE: The information in this section is specific to the knative branch of Dispatch. Equivalent documentation for Dispatch Solo can be found on the
solo
branch.
The diagram below illustrates the different components which make up the Dispatch project:
NOTE: The information in this section is specific to the knative branch of Dispatch. Equivalent documentation for Dispatch Solo can be found on the
solo
branch.
-
export GCLOUD_KEY=<path to key.json>
-
Create GKE cluster:
K8S_VERSION=1.10.7-gke.6 export CLUSTER_NAME=dispatch-knative gcloud container clusters create -m n1-standard-4 --cluster-version ${K8S_VERSION} ${CLUSTER_NAME} gcloud container clusters get-credentials ${CLUSTER_NAME}
-
Install Knative:
# Get the current knative verision used with dispatch KNATIVE_VERSION=$(cat Gopkg.toml | grep -A 2 'name = "github.com/knative/serving"' | grep revision | cut -d '"' -f2) ./scripts/install-knative.py ${CLUSTER_NAME} --gcloud-key=${GCLOUD_KEY} --revision=${KNATIVE_VERSION}
In order to install Knative, follow the development instructions
Installing Dispatch depends on having a Kubernetes cluster with the Knative components installed (Build, Serving and soon Eventing). From here build and install dispatch as follows:
-
Set the following environment variables:
export DISPATCH_NAMESPACE="dispatch-server" export DISPATCH_DEBUG="true" export RELEASE_NAME="dispatch-server" export MINIO_USERNAME="dispatch" export MINIO_PASSWORD="dispatch" export INGRESS_IP=$(kubectl get service -n istio-system knative-ingressgateway -o json | jq -r .status.loadBalancer.ingress[].ip)
-
Build and publish a dispatch image:
PUSH_IMAGES=1 make images
-
The previous command will output a configuration file
values.yaml
:image: host: username tag: v0.1.xx storage: minio: username: ******** password: ********
-
Deploy via helm chart (if helm is not installed and initialized, do that first):
helm upgrade -i --debug ${RELEASE_NAME} ./charts/dispatch --namespace ${DISPATCH_NAMESPACE} -f values.yaml
NOTE: Use following to create cluster role binding for tiller:
kubectl create clusterrolebinding tiller-cluster-admin --clusterrole=cluster-admin --serviceaccount=kube-system:default
-
Reconfigure Knative serving (need to whitelist our internal repository):
./scripts/configure-knative.sh
-
Build the CLI (substitute darwin for linux if needed):
make cli-darwin # Create symlink to binary ln -s `pwd`/bin/dispatch-darwin /usr/local/bin/dispatch
-
Create the Dispatch config:
cat << EOF > config.json { "current": "${RELEASE_NAME}", "contexts": { "${RELEASE_NAME}": { "host": "$(kubectl -n ${DISPATCH_NAMESPACE} get service ${RELEASE_NAME}-nginx-ingress-controller -o json | jq -r .status.loadBalancer.ingress[].ip)", "port": 443, "scheme": "https", "insecure": true } } } EOF # point to the config file (could also move to ~/.dispatch/config) export DISPATCH_CONFIG=`pwd`/config.json
-
Test out your install: First, create an baseimage:
dispatch create baseimage python3-base dispatchframework/python3-base:0.0.13-knative Created baseimage: python3-base
Then, create an image:
dispatch create image python3 python3-base Created image: python3
Wait for status READY:
dispatch get images NAME | DESTINATION | BASEIMAGE | STATUS | CREATED DATE -------------------------------------------------------------------------- python3 | *********** | ********* | READY | Tue Sep 25 16:51:35 PDT 2018
Create a function:
dispatch create function --image python3 hello ./examples/python3/hello.py Created function: hello
Once status is READY:
dispatch get function NAME | FUNCTIONIMAGE | STATUS | CREATED DATE ---------------------------------------------------------------- hello | ************* | READY | Thu Sep 13 12:41:07 PDT 2018
Exec the function:
dispatch exec hello <<< '{"name": "user"}' | jq . { "context": { "logs": { "stdout": [ "messages to stdout show up in logs" ], "stderr": null } }, "payload": { "myField": "Hello, user from Nowhere" } }
Create an endpoint:
dispatch create endpoint get-hello hello --method GET --method POST --path /hello
Hit the endpoint with curl:
curl -v http://${INGRESS_IP}/hello?name=Jon -H 'Host: default.dispatch-server.dispatch.local'
For a more complete quickstart see the developer documentation