razeedash-api

Razeedash-API is the interface used by

• app.razee.io
• watch-keeper

Requirements

• Kubernetes CLI Tools
• Kubernetes Cluster
• MongoDB

Environment Variables

| Name | Required | Default Value | | ---- | -------- | ------------- | | MONGO_URL | yes | 'mongodb://localhost:3001/meteor' | | MONGO_DB_NAME | yes | 'meteor' | | S3_ENDPOINT | no | n/a | | S3_ACCESS_KEY_ID | if S3_ENDPOINT defined | n/a | | S3_SECRET_ACCESS_KEY | if S3_ENDPOINT defined | n/a | | S3_LOCATION_CONSTRAINT | no | 'us-standard'| | S3_CHANNEL_BUCKET | no | 'razee'| | S3_RESOURCE_BUCKET | no | S3_CHANNEL_BUCKET or 'razee'| | ORG_ADMIN_KEY | no | n/a | | ADD_CLUSTER_WEBHOOK_URL | no | n/a | | AUTH_MODEL | no | 'default' [default, local, passport.local] are supported |

If S3_ENDPOINT is defined then encrypted cluster YAML is stored in S3 otherwise it will be stored in the mongoDB.

ORG_ADMIN_KEY is required if you plan on adding organizations using the api/v2/orgs endpoint

ADD_CLUSTER_WEBHOOK_URL signifies the webhook endpoint to hit when a cluster is added. Razee will do a POST request to this url with json data { org_id, cluster_id, cluster_name }. If a razeedash-add-cluster-webhook-headers-secret exists in the namespace, its key-value pairs will be used as headers in the request. For instance, if you would like to send an Authorization header in the request to verify that razee is sending the webhook, you can create a secret like so:

apiVersion: v1
kind: Secret
metadata:
  namespace: razee
  name: razeedash-add-cluster-webhook-headers-secret
stringData:
  Authorization: SOME_APIKEY

For local development, put the headers as files in the /var/run/secrets/razeeio/razeedash-api/add-cluster-webhook-headers directory.
For instance:
echo "SOME_APIKEY" > /var/run/secrets/razeeio/razeedash-api/add-cluster-webhook-headers/Authorization
(you may need sudo to perform this operation).

OS/X

gettext package is default on most Linux systems. If you are using OS/X for local development you may need to install it in order to generate a deployment YAML.

If you are testing ./build/process-template.sh you will need brew installed and gettext.

brew update
brew install gettext
brew link --force gettext

Install on Kubernetes

Setup so you can use kubectl commands on the target cluster. For IBM Cloud Kubernetes Service the following command will get the KUBECONFIG for your Kubernetes cluster and export the KUBECONFIG variable.

ibmcloud ks cluster-config <cluster name>

Create secrets and deploy

Generate a base64 encoding for the mongo_url to be used in the razeedash-secret. The following is an example of local mongo deployment. Not recommended for production use.

echo -n "mongodb://mongo:27017" | base64

Note: Production MongoDB usually is a minimum of 3 nodes using replica sets. That definition would look something like:

echo -n "mongodb://mongo‑0:27017,mongo‑1:27017,mongo‑2/razeedash?replicaSet=rs0&tls=true" | base64

tls=true should be at the end of your connection string when connecting to a hosted mongo.

Create file razeedash-secret.yaml using the generated string provided from the previous command.

apiVersion: v1
kind: Secret
metadata:
  name: razeedash-secret
  namespace: razee
type: Opaque
data:
  mongo_url: bW9uZ29kYjovL21vbmdvOjI3MDE3L3JhemVlZGFzaAo=

Add org_admin_key to the data section of razeedash-secret in order to control organizations using the api/v2/orgs endpoint

echo -n abcdefghijklmnop012345678 | base64
# outputs YWJjZGVmZ2hpamtsbW5vcDAxMjM0NTY3OA==

apiVersion: v1
kind: Secret
metadata:
  name: razeedash-secret
  namespace: razee
type: Opaque
data:
  mongo_url: bW9uZ29kYjovL21vbmdvOjI3MDE3L3JhemVlZGFzaAo=
  org_admin_key: YWJjZGVmZ2hpamtsbW5vcDAxMjM0NTY3OA==

If you are using your own managed mongodb system, make sure you setup the mongo_url secret properly. For example, your mongo_url connection string might look something like this:

echo -n "mongodb://mongo‑0:27017,mongo‑1:27017,mongo‑2/razeedash?replicaSet=rs0&tls=true" | base64
# bW9uZ29kYjovL21vbmdv4oCRMDoyNzAxNyxtb25nb+KAkTE6MjcwMTcsbW9uZ2/igJEyL3JhemVlZGFzaD9yZXBsaWNhU2V0PXJzMCZ0bHM9dHJ1ZQ==

Note that tls=true should be at the end of your connection string.

You will also need to add mongo_cert to razeedash-secret. This will contain a base64 encoded copy of the tls certificate used to access your managed mongodb.

apiVersion: v1
kind: Secret
metadata:
  name: razeedash-secret
  namespace: razee
type: Opaque
data:
  mongo_url: bW9uZ29kYjovL21vbmdv4oCRMDoyNzAxNyxtb25nb+KAkTE6MjcwMTcsbW9uZ2/igJEyL3JhemVlZGFzaD9yZXBsaWNhU2V0PXJzMCZ0bHM9dHJ1ZQ==
  mongo_cert: LS0tLS1CRUdJTiBDRVJUSUZJQ0FURS0tLS0tCnlvdXIgbW9uZ28gY2VydCBnb2VzIGhlcmUKLS0tLS1FTkQgQ0VSVElGSUNBVEUtLS0tLQo=

Apply the secret to kubernetes, build the resource.yml and apply to cluster:

kubectl apply -f razeedash-secret.yaml
./build/process-template.sh kubernetes/razeedash-api/resource.yaml >/tmp/resource.yaml
kubectl apply -f /tmp/resource.yaml

Check logs on all deployed pods to make sure there are no errors.

for i in `kubectl get pods -n razee --selector=app=razeedash-api | \
  grep razeedash-api | \
  awk '{print $1}'`; do kubectl logs ${i} -n razee --since 5m; done

Example deployment using IBM Cloud

This will deploy the razeedash-api and mongo on a 3 node cluster using IBM Cloud Kubernetes Service.

Note: In a production scenario it is recommended to used a managed Mongo database service, like IBM Cloud Databases for MongoDB.

Requirements:

• jq jq is a lightweight and flexible command-line JSON processor
• IBM Cloud Account IBM Cloud
• IBM Cloud CLI Setting up the CLI and API
• Kubernetes CLI Install and Set Up kubectl

Create Cluster

You can use a utility script ic_create_cluster.sh located in kube-cloud-scripts or follow the (IBM Containers CLI plugin documentation to create a cluster.

ic_create_cluster.sh --name razeetest

if you have an existing cluster and need to resize

ibmcloud ks worker-pool-resize \
  --cluster <cluster-name> \
  --worker-pool default \
  --size-per-zone 3

Once the cluster (ibmcloud ks clusters) is created and in a normal state, we need to get Kubernetes config.

ibmcloud ks cluster-config razeetest

Example

ibmcloud ks cluster-config razeetest
OK
The configuration for razeetest was downloaded successfully.

Export environment variables to start using Kubernetes.

export KUBECONFIG=~/.bluemix/plugins/container-service/clusters/razeetest/kube-config-wdc07-razeetest.yml

Note: Setup 3 node MongoDB Cluster must have a minimum of 3 nodes in order to statisfy Mongo. You can follow the guide Setting up clusters and workers to deploy a 3 node MongoDB replica set.

Deploy components

Deploy MongoDB and set up replica sets. This is based on the guide Deploy a MongoDB replica set using IBM Cloud Kubernetes Service Individually

# Add razee namespace, single mongo, razeedash secret
kubectl apply -f samples/namespace.yaml
kubectl apply -f samples/persistentVolume.yaml
kubectl apply -f samples/persistentVolumeClaim.yaml
kubectl apply -f samples/mongo.yaml
kubectl apply -f samples/service.yaml
kubectl apply -f samples/secret.yaml

or All in one command

# Add razee namespace, single mongo, razeedash secret
kubectl apply -f samples/allinone.yaml

Wait until mongo pods are ready. You can check the status via:

kubectl get pods

Once pods are in a Running state continue with the setup process

# Get latest release of razeedash-api and deploy
kubectl apply -f "https://github.com/razee-io/razeedash-api/releases/latest/download/resource.yaml"

Check logs across pods using kc_logs.sh script from kube-cloud-scripts

kc_logs.sh razee razeedash-api 1m

Swagger API

Swagger UI is available and if started locally can be accessed via the following URL: http://localhost:3333/api-docs/

Web hooks

Implemented web hooks so data in Razeedash can be augmented by third-party services such as test suites and vulnerability scanners.

Components are:

• Web hook creation
• Web hook deletion
• Trigger by resource ID
• Trigger by Regex
• Callback response

Web hook definition

POST /v2/webhook/ will create a web hook and the header must contain the razee-org-key

JSON body for web hook triggered by cluster change:

{
          "cluster_id": "ID of the cluster",
          "trigger": "cluster",
          "kind": "Deployment",
          "field": "searchableData.name",
          "filter": "regex string example to match, `(watch-keeper)`",
          "service_url": "URL of service to POST upon triggering"
}

Note: field and filter are optional. If not defined the above will fire if a deployment resource kind is changed

JSON body for web hook triggered by image change:

{
          "trigger": "image",
          "kind": "image",
          "field": "name",
          "filter": "regex string example, `(quay.io\\/mynamespace)`",
          "service_url": "URL of service to POST upon triggering"
}

Note: field and filter are optional. If not defined the above will fire if any new image is deployed on an organization's clusters

• kind: Kind would be the type of resource or in a special case, images
• trigger: cluster, image
• id: ID of the resource used in trigger OR
• field: field dot-notation into JSON to apply a filter
• filter: regex parameter to match if the trigger should fire or not if field is defined
• service URL: URL to call if the web hook is triggered

Web hook deletion

DELETE /v2/webhook/:id will delete the web hook. Current badge data will not be affected by the deletion. If a callback for that webhook_id occurs then the badge would be removed from the resource and 404 sent back to the calling service.

Trigger Logic

Trigger points will be added to Razeedash API for when:

• New image is deployed
• A resource kind is deployed or modified on a specific cluster

Trigger will also have to pass regex filter in order to fire the web hook. If determine to call the web hook the service URL is called and relevant data posted along with a callback URL. The called service can augment the data by calling the callback URL.

Razeedash API will look up the resource to make sure it is still being used and if not, return a 404 indicating to the calling service it should no longer provide updates on this web hook. If found the badge information will added to resource.

Calling the service by cluster trigger POSTs:

{
  "cluster_id": "ID of the cluster",
  "cluster_metadata": "JSON array for the cluster metadata",
  "resource_id": "ID of the resource to badge",
  "resource_kind": "What kind of resource",
  "resource": "JSON of the resource object",
  "webhook_id": "ID of the webhook definition",
  "callback_url":  "URL to POST badge data"
}

Calling the service by image trigger POSTs:

{
  "image_id": "ID of the image",
  "image_name": "Image name",
  "webhook_id": "ID of the webhook definition",
  "callback_url":  "URL to POST badge data"
}

Scenario: Trigger by resource ID

User defines a web hook that if a deployment changes or something new is added to call a web hook to run integration tests.

{
  "kind": "Deployment",
  "trigger": "cluster",
  "id": "fb56c61b676844d292f1f18e719c31f2",
  "service_url": "https://my.testingservice.com/run_integration"
}

New deployment is rolled out to cluster designated as the "staging" cluster environment. Web hook is called to service_url and the deployment.yaml, org_id, cluster_id are all posted along with a a call back URL.

The testing service calls the callback with badge data:

• badge: URL of a running man
• description: "Running tests"
• link: (link to the live tests being run)
• status: info

Testing service runs integration tests on the staging cluster and and calls the callback with badge data:

• badge: URL of a green circle
• description: "All tests completed successfully"
• link: (link to the test logs)
• status: info

Scenario: Trigger by regex

User defines a web hook that if a deployment changes or something new is added to call a web hook to run integration tests.

{
  "kind": "image",
  "trigger": "image",
  "field": "name",
  "filter": "(quay.io\/mynamespace)",
  "service_url": "https://my.quayscanner.com/check"
}

When a new image is deployed, the name of the image is checked against the filter, if defined, and then the service_url is called with image name, image ID, org_id and callback URL.

The scanner service calls the callback with badge data:

• badge: URL image of binoculars
• description: "Looking for vulnerabilities"
• link: (link to service of that image being checked)
• status: info

Razeedash API checks to see if the image is still in use and returns a 201

The image is scanned and shown clean and will be rechecked by the service in 24 hours. In the meantime it calls the callback with badge data:

• badge: URL image of green circle
• description: "no vulnerabilities"
• link: (link to service of that image results)
• status: info

Razeedash API checks to see if the image is still in use and returns a 201

24 hours later a minor vulnerabilities is discovered and the service calls the callback again with badge data:

• badge: URL image of yellow circle
• description: "minor vulnerabilities detected"
• link: (link to service of that image results)
• status: warning

Razeedash API checks to see if the image is still in use and returns a 201

24 hours later a major vulnerability is discovered and the service calls the callback again with badge data:

• badge: URL image of red circle
• description: "major vulnerabilities detected"
• link: (link to service of that image results)
• status: error

Razeedash API checks to see if the image is still in use and finds it is not and returns a 404 to the vulnerability service. The vulnerability service should then stop reporting that image security issues from now on.

Scenario: Filter a specific resource

User defines a web hook that filters for a specific resource. In this case we are looking for new or changed deployments where the field metadata.name matches watch-keeper on a specific cluster to trigger the web hook.

{
  "kind": "Deployment",
  "field": "metadata.name",
  "filter": "(watch-keeper)",
  "trigger": "cluster",
  "id": "fb56c61b676844d292f1f18e719c31f2",
  "service_url": "https://my.testingservice.com/run_integration"
}

New deployment is rolled out to cluster designated as the "staging" cluster environment. Web hook is called to service_url and the deployment.yaml, org_id, cluster_id are all posted along with a a call back URL.

Callback response from remote service

When the remote service wants to add a badge as a result of the web hook call, the POST to /v2/callback should have the following body:

Header should contain the razee-org-key

{
    "webhook_id": "id of the webhook from initial call to service",
    "url": "URL of the badge image",
    "description": "short description of badge",
    "link": "URL link for details",
    "status": "info | error | warning"
}

Razeedash API will accept the callback URL and make sure the webhook and resource is still valid. If the resource is no longer in use or the webhook was deleted then the callback response will return a 404. If valid it will add the augmented data to the resource.

The resource will have a new attribute badges. The badge will replace any existing badge with the same webhook_id or if it does not exist, add to the array.

GraphQL for local development

export AUTH_MODEL=local

Then start the razeedash-api server, you will see a message like bellow from the console

🏄 Apollo server listening on http://[::]:3333/graphql

the graphql playground is enabled and could be accessed at http://localhost:3333/graphql
if NODE_ENV is not equal to production. For local authorization model, signUp graphql
API is provided to sign-up a user, for example:

mutation {
  signUp(
    username: "[email protected]"
    email: "[email protected]"
    password: "password123"
    org_name: "test_org"
    role: "ADMIN"
  ) {
    token
  }
}

If a user is already signed up, then signIn api could be used to sign-in a user, for example:

mutation {
  signIn(login: "[email protected]" password:"password123") {
    token
  }
}

Both APIs return a JWT token, which you could use to query other graphql APIs. e.g. Following graphql query, will return organizations a user belongs to:

query {organizations {
  _id
  name
}}

With the following HTTP Header:

{"Authorization": "Bearer <the token value returned from signUp or signIn>"}

You could also query registrationUrl for the user, e.g.

query {
  registrationUrl(org_id: "<the orgnization_id returned from organizations graphql api >") {
    url
  }
}

With the following HTTP Header:

{"Authorization": "Bearer <the token value returned from signUp or signIn>"}

For all other supported graphql APIs, please click DOCS or SCHEMA from the graphql play-ground.