Kiali

Introduction

kiali is a management console for Istio service mesh. Kiali can be quickly installed as an Istio add-on or integrated as a trusted component within a production environment.

Introduction
Developer setup
Configuration
Embedding Kiali
Configure External Services
- Grafana
Additional Notes
Exposing Kiali to External Clients Using Istio Gateway
Experimental
- Observing a Remote Cluster

Contributing

First, check the Community section on kiali.io, which provides a brief introduction on contributing, how to report issues and request features, and how to reach us.

For those interested in making code contributions, please also check the Contribution Guide.

Getting Started

The target audience of this README are developers. For those who are not developers but wish to learn more about Kiali, the Kiali documentation should be more helpful. For instructions on installing Kiali, please read the Installation page.

How and where Kiali is released?

Read the RELEASING.adoc file.

Developer setup

Ensure the following tools are available:

The Go Programming Language
- Kiali releases are built with a specified minimum version of Go, as indicated in the Makefile. While Kiali may compile with other versions, using the version specified in the Makefile is recommended for consistent builds.
git
gcc
Docker or Podman
- If using podman, set the environment variable DORP=podman.
NodeJS (Node.js >= 20 with the NPM command)
Yarn
The GNU make (or a compatible alternative)

Once the required developer tools are available, the code can be obtained and built using the following script:

# Checkout the source code
mkdir kiali_sources
cd kiali_sources
export KIALI_SOURCES=$(pwd)

git clone https://github.com/kiali/kiali.git
git clone https://github.com/kiali/kiali-operator.git
git clone https://github.com/kiali/helm-charts.git

ln -s $KIALI_SOURCES/kiali-operator kiali/operator

# Build the back-end and run the tests
cd $KIALI_SOURCES/kiali
make build test

# Go test flags can be passed via the GO_TEST_FLAGS environment variable.
# make -e GO_TEST_FLAGS="-race -v -run=\"TestCanConnectToIstiodReachable\"" test

# Build the front-end and run the tests
make build-ui-test

Note	The rest of this README assumes the directory tree created by the previous commands:

-- kiali_sources
   |- kiali
   |- kiali-operator
   \- helm-charts

Create a Kubernetes cluster and install a Service Mesh

To use Kiali, a management console for Istio-based service meshes, an Istio-compatible service mesh is required. Istio meshes are installed on Kubernetes clusters.

We provide a few unsupported scripts that can help to get started. The scripts below are already in the Kiali Repository. Ensure that the correct path is being used to run the scripts without issues.

The crc-openshift.sh script allows creation of an OpenShift cluster on a local machine.
For those familiar with minikube, the k8s-minikube.sh script can be used. This script includes an option to install Dex, which is helpful for testing with OpenID.
Alternatively, the start-kind.sh script can be used to create a single-node KinD cluster with MetalLB enabled, allowing Kiali to be tested in a real environment.
Finally, the install-istio-via-istioctl.sh and the install-bookinfo-demo.sh scripts can assist in installing Istio and the Bookinfo sample application in the cluster, respectively. Trying to run these scripts without any arguments is also an option.

These scripts are written to rely on the minimum dependencies as possible and will try to download any required tools.

Based on the type of cluster in use, define the CLUSTER_TYPE environment variable in the shell. Set it to openshift (default if not specified), minikube or kind so the Makefiles can assist with additional operations. For other cluster types, set CLUSTER_TYPE=local.

Note	For `minikube` it is recommended to enable the `registry` and `ingress` add-ons, which can be configured by the `k8s-minikube.sh` script.

Note	When using `docker` with Minikube’s registry add-on or any custom non-secure registry, ensure the Docker daemon is properly configured to use your registry.

Building the Container Image and deploying to a cluster

Assuming the following conditions:

The back-end and front-end have been successfully built
A Kubernetes cluster with an Istio-based Service Mesh has been created
The CLUSTER_TYPE=local environment variable is not being used

The commands below will deploy a development build of Kiali to the cluster.

cd $KIALI_SOURCES/kiali

# Build the Kiali-server and Kiali-operator container images and push them to the cluster
make cluster-push

# To build and push only the Kiali-server container images:
# make cluster-push-kiali

# To build and push only the Kiali-operator container images:
# make cluster-push-operator

# Deploy the operator to the cluster
make operator-create

# Create a KialCR to instruct the operator to deploy Kiali
make kiali-create

When using the CLUSTER_TYPE=local environment variable, additional environment variables must be declared to specify the container registry for pushing container images. The make container-push* targets should be used instead of the cluster-push* targets. For example, if the container registry is localhost:5000:

export QUAY_NAME=localhost:5000/kiali/kiali
export CONTAINER_NAME=localhost:5000/kiali/kiali
export OPERATOR_QUAY_NAME=localhost:5000/kiali/kiali-operator
export OPERATOR_CONTAINER_NAME=localhost:5000/kiali/kiali-operator

cd $KIALI_SOURCES/kiali

# Build the Kiali-server and Kiali-operator container images and push them to the cluster
make container-build container-push

# To build and push only the Kiali-server container images:
# make container-build-kiali container-push-kiali-quay

# To build and push only the Kiali-operator container images:
# make container-build-operator container-push-operator-quay

# Deploy the operator to the cluster
make operator-create

# Create a KialCR to instruct the operator to deploy Kiali
make kiali-create

Reloading Kiali image

If Kiali is already installed and there is a need to recreate the Kiali server pod, the following command can be run:

cd $KIALI_SOURCES/kiali
make kiali-reload-image

This is intended to facilitate development. To quickly build a new Kiali container image and load it to the cluster, run the following command:

cd $KIALI_SOURCES/kiali/frontend
yarn && yarn build

cd $KIALI_SOURCES/kiali
make clean build cluster-push-kiali kiali-reload-image

Note	No equivalent reload command exists for the operator. The operator must be manually reloaded using `kubectl` or `oc` commands.

Cluster clean-up

cd $KIALI_SOURCES/kiali

# Delete the Kiali CR to let the operator remove Kiali.
make kiali-delete

# If the previous command fails to complete, the following command forces removal by bypassing the operator
# make kiali-purge

# Remove the operator
# NOTE: After this is completed, the `kiali-create` and `kiali-delete` targets will be ineffective
# Until the `operator-create` target is run to re-deploy the Kiali operator, execute the following command to delete the operator:
make operator-delete

Code formatting and linting

When changing the back-end code of Kiali, ensure that the changes are properly formatted and no new linting issues are introduced before submitting a pull request by running:

# CD to the back-end source code
cd $KIALI_SOURCES/kiali

# Install linting tools
make lint-install

# Format the code and run linters
make format lint

Enable tracing

Kiali itself is instrumented with opentelemetry tracing to help provide insights and surface performance issues for the kiali server. To enable, set the server.observability.tracing.enabled and server.observability.tracing.collector_url configuration options.

apiVersion: kiali.io/v1alpha1
kind: Kiali
metadata:
  name: kiali
spec:
...
  server:
    observability:
      tracing:
        collector_url: http://jaeger-collector.istio-system:14268/api/traces
        enabled: true
...

Running Standalone

For debugging purposes, running Kiali outside of a cluster environment may be desired. To do this, use the run-kiali.sh script in the hack directory. Check the --help output for available options. By default, the script uses the configuration template file located in the same directory it uses, and can be found in the config template file also located in the hack directory. For additional information, refer to the comments at the top of both files.

cd $KIALI_SOURCES/kiali/hack
./run-kiali.sh

Running integration tests

There are two sets of integration tests. The first are backend tests that test the Kiali API directly. These can be found at backend tests. The second are frontend Cypress tests that test Kiali through the browser. These can be found at frontend tests.

Both tests are executed as part of the CI pipeline. To run these tests locally, the script can be used to setup a local environment and execute the integration tests. Alternatively, the tests can be run against any live environment that meets the following requirements.

Requirements: - Istio - Kiali - bookinfo demo app - error rates demo app

This script can be used install all necessary demo applications for testing and supports deployments on both Openshift and non-Openshift enviroments.

# For frontend development, start the frontend development server, where `<kiali-url>` is the URL to the base Kiali UI location, such as `http://localhost:20001/kiali`:
make -e YARN_START_URL=http://<kiali-url> yarn-start

# Start the cypress tests. The tests will run against the frontend development server by default.
# Alternatively, a custom URL can be provided using environment variables:
#
# make -e CYPRESS_BASE_URL=http://<kiali-url> cypress-gui
make cypress-gui

Note that make cypress-gui runs the Cypress GUI, enabling the selection of individual tests to run. To execute the entire test suite in headless mode, use the cypress-run target instead.

Debugging Server Backend

VisualStudio Code

For VisualStudio Code, install the following launcher.json to launch the Kiali Server in the debugger. First, run the hack/run-kiali.sh script first to ensure that required services, such as Prometheus port-forward proxy, are running.

{
    // To use this, first run "hack/run-kiali.sh --tmp-root-dir $HOME/tmp --enable-server false"
    // Pass in --help to that hack script for details on more options.
    "version": "0.2.0",
    "configurations": [
        {
            "name": "Launch Kiali to use hack script services",
            "type": "go",
            "request": "launch",
            "mode": "debug",
            "program": "${workspaceRoot}/kiali.go",
            "cwd": "${env:HOME}/tmp/run-kiali",
            "args": ["-config", "${env:HOME}/tmp/run-kiali/run-kiali-config.yaml"],
            "env": {
                "KUBERNETES_SERVICE_HOST": "127.0.0.1",
                "KUBERNETES_SERVICE_PORT": "8001",
                "LOG_LEVEL": "trace"
            }
        }
    ]
}

Debugging GUI Frontend

The Kiali GUI can be debugged directly in Google Chrome using Chrome Developer Tools or within Visual Studio Code.

To use either, some initial steps are required.

First, start the Kiali Server backend, either within a cluster or by using the run-kiali.sh script as previously explained. Before proceeding, determine the Kiali URL. For example, when Kiali is run in Minikube with a port-forward set up to expose it, the URL to note will be`http://localhost:20001/kiali`.

Next, run the GUI frontend using the following command: make -e YARN_START_URL=<kiali-url> yarn-start where <kiali-url> is the URL determined in the previous step. Additionally, if needed, pass -e PORT=3001 to override the default port 3000, which may conflict with Grafana if the Kiali Server was started using run-kiali.sh. Some examples:

If the Kiali Server is running in minikube with a port-forward exposing it, then run make -e YARN_START_URL=http://localhost:20001/kiali yarn-start.
If the Kiali Server is running in OpenShift with the usual Kiali Route exposing it, then run make -e YARN_START_URL=https://<Kiali-OpenShift-Route-IP>/ yarn-start.
If the Kiali Server is running locally via run-kiali.sh, then run make -e YARN_START_URL=http://localhost:20001/kiali -e PORT=3001 yarn-start.

The yarn-start make command will start the Kiali GUI frontend on a local endpoint. Once ready, check the output for the "Local" URL to access it. The output will resemble the following:

Compiled successfully!

The @kiali/kiali-ui can now be viewed in the browser.

  Local:            http://localhost:3001
  On Your Network:  http://192.168.1.15:3001
...

At this point, the debugger tool of choice can be set up - refer to the following sections for details.

Google Chrome Developer Tools

Start Google Chrome and point the browser to the local URL for the Kiali GUI frontend started by yarn-start (in the example above, that will be http://localhost:3001).

In Google Chrome, open the Developer Tools. Press F12 or Control-Shift-I to do this.

Within the Developer Tools, navigate to the Sources tab, then the Filesystem sub-tab, and press the + Add folder to workspace link. In the file selection dialog, select the Kiali frontend/src folder. This will inform Developer Tools where the Kiali GUI frontend source code can be found.

At this point, Google Chrome requires permission to access the local source code folder. A prompt will appear at the top of the browser window - click the "Allow" button to grant Chrome the necessary permissions.

The Kiali Server frontend is now ready for debugging. Breakpoints can be set, variables inspected, stack traces examined, and other typical debugging actions can be performed. ==== VisualStudio Code

For VisualStudio Code, the following launcher.json can be installed to launch Google Chrome for debugging the Kiali Server GUI frontend. The url setting should correspond to the local URL of the yarn-start server - ensuring the correct URL is used for the specific environment.

{
    "version": "0.2.0",
    "configurations": [
        {
            "name": "Launch Chrome",
            "type": "chrome",
            "request": "launch",
            "url": "http://localhost:3001",
            "webRoot": "${workspaceFolder}"
        }
    ]
}

Configuration

Many configuration settings can optionally be set within the Kiali Operator custom resource (CR) file. See this example Kiali CR file that has all the configuration settings documented.

Embedding Kiali

To embed Kiali in other applications, Kiali provides a feature called Kiosk mode. In this mode, the main header and navigation bar are not displayed.

To enable Kiosk mode, simply add a kiosk=<platform_id> URL parameter. The full path of the page to be embedded must be used. For example, if Kiali is accessed via HTTPS:

To embed the Overview page, use https://kiali_path/overview?kiosk=console.
To embed the Graph page, use https://kiali_path/graph/namespaces?kiosk=console.
To embed the Applications list page, use https://kiali_path/applications?kiosk=console.

If the page to be embedded includes other URL arguments, any of them can be specified to preset options. For example, to embed the graph of the bookinfo namespace, use the following URL: http://kiali_path/graph/namespaces?namespaces=bookinfo&kiosk=console.

<platform_id> value in the kiosk URL parameter will be used in future use cases to add conditional logic on embedded use cases, for now, any non empty value will enable the kiosk mode.

Configure External Services

Grafana

If Grafana is installed in a custom manner that Kiali cannot automatically detect, the value of grafana > url in the Kiali CR must be updated.

apiVersion: kiali.io/v1alpha1
kind: Kiali
metadata:
  name: kiali
spec:
...
    external_services:
      grafana:
        url: http://grafana-istio-system.127.0.0.1.nip.io
...

Additional Notes

Frontend development guidelines

Frontend development guidelines (styles, i18n, etc.) can be found here

Upgrading Go

The Kiali project will periodically upgrade to a newer version of Go. These are the steps that need to be performed in order for the Kiali build to use a different version of Go:

Run go mod edit -go=x.y where "x" and "y" are the major/minor versions of the Go version being used.
Run go mod tidy -v
Run make clean build build-ui test to ensure everything builds correctly. If any problems occur, fix them.
Commit the changes to the working branch, create a PR, and make sure everything builds and works before merging the PR.

The Makefile and some Github Actions will check the go version from the go.mod file.

Procedure to check and update patternfly versions

Launch command npx npm-check-updates -t latest -f '/^@patternfly/'
Launch yarn install to update the yarn.lock
Add to the commit package.json and yarn.lock

Running the UI outside the cluster

When developing the Kiali UI running it outside of the cluster can be helpful for easily updating the UI code and viewing changes without needing to re-deploy. The preferred method for this is to use React’s proxy feature, as outlined here. Alternatively, the make -e YARN_START_URL=<url> yarn-start command can be used, where <url> points to the Kiali backend.

Disabling SSL

In the provided OpenShift templates, SSL is enabled by default. To disable it, the following steps should be taken:

Remove the "tls: termination: reencrypt" option from the Kiali route
Remove the "identity" block, with certificate paths, from the Kiali Config Map.
Optionally, the annotation "service.beta.openshift.io/serving-cert-secret-name" can be removed from the Kiali Service, along with the related kiali-cabundle volume declared and mounted in Kiali Deployment. However, if not removed, these will simply be ignored.

Exposing Kiali to External Clients Using Istio Gateway

The operator will create a Route or Ingress by default (see the Kiali CR setting "deployment.ingress_enabled"). To expose Kiali via Istio, Gateway, Virtual Service, and Destination Rule resources can be created, as shown below:

---
apiVersion: networking.istio.io/v1
kind: Gateway
metadata:
  name: kiali-gateway
  namespace: istio-system
spec:
  selector:
    istio: ingressgateway
  servers:
  - port:
      number: 80
      name: http-kiali
      protocol: HTTP
    # https://istio.io/latest/docs/reference/config/networking/gateway/#ServerTLSSettings
    tls:
      httpsRedirect: false
    hosts: [<your-host>]
  - port:
      number: 443
      name: https-kiali
      protocol: HTTPS
    tls: {}
    hosts: [<your-host>]
...
---
apiVersion: networking.istio.io/v1
kind: VirtualService
metadata:
  name: kiali-virtualservice
  namespace: istio-system
spec:
  gateways:
  - kiali-gateway
  hosts: [<your-host>]
  http:
  - route:
    - destination:
        host: kiali.istio-system.svc.cluster.local
        port:
          number: 20001
      weight: 100
...
---
apiVersion: networking.istio.io/v1
kind: DestinationRule
metadata:
  name: kiali-destinationrule
  namespace: istio-system
spec:
  host: kiali
  trafficPolicy:
    tls:
      mode: DISABLE
...

Experimental

Observing a Remote Cluster

Note	The "Central IstioD" setup is currently named "Primary-remote" multi-cluster setup.

Warning

When this support was incorporated into Kiali, the "Central IstioD" setup of Istio was in an early development phase. These instructions are probably now broken.

There are certain use cases where Kiali needs to be deployed in one cluster (Control Plane) and observe a different cluster (Data Plane).

Follow these steps:

1: Have the Istio with an External Control Plane setup running

2: Create the Kiali ClusterRole, ClusterRoleBinding, and ServiceAccount in the Data Plane cluster

3: Create a remote secret in the Control Plane, using the Data Plane ServiceAccount that was just created. This allows the Control Plane to read from and modify the Data Plane

istioctl create-remote-secret --service-account kiali-service-account --context=$DataPlane --name kiali | kubectl apply -n istio-system --context=$ControlPlane -f -

4: Kiali will now run in the Control Plane. The remote secret must be added to the Kiali Deployment by specifying a Volume and VolumeMount. When Kiali detects /kiali-remote-secret/kiali it will use the remote cluster’s API server instead of the local API server

spec:
  template:
    spec:
      containers:
      - volumeMounts:
        - mountPath: /kiali-remote-secret
          name: kiali-remote-secret
      volumes:
      - name: kiali-remote-secret
        secret:
          defaultMode: 420
          optional: true
          secretName: istio-remote-secret-kiali

5: Kiali now requires Istio metrics from the sidecars. Prometheus must be run in the Control Plane and configured to scrape the metrics from an envoyMetricsService. These metrics are required.

6: Kiali in the Control Plane should now be fully functional with the Data Plane

kiali / kiali