This is a repository for Tyk helm charts. Please visit the official website to learn about Tyk's licensing and deployment options and the different Helm Charts available for Kubernetes deployment.
Helm must be installed to use the charts. Please refer to Helm's documentation to get started.
Once Helm is set up properly, add the repo as follows:
helm repo add tyk-helm https://helm.tyk.io/public/helm/charts/You can then run helm search repo tyk-helm to see the charts.
Helm umbrella chart (chart of charts) is an easy and really flexible way of installing multiple components as a single one. We have following umbrella charts that help you to install group of related tyk components based on your deployment need.
| Umbrella Charts | Description | Status |
|---|---|---|
| tyk-oss | Tyk Open Source | Stable |
| tyk-stack | Tyk Self Managed | Stable |
| tyk-control-plane | Tyk Self Managed (MDCB) Control Plane | Stable |
| tyk-data-plane | Tyk Self Managed (MDCB) Data Plane Tyk Hybrid Data Plane |
Stable |
The Tyk Operator chart is in components/tyk-operator and is published as tyk-helm/tyk-operator. Its CRDs are in tyk-operator-crds. The tyk-operator repository is archived. To install Tyk Operator, see Install Tyk Operator.
- Redis is required for all of the Tyk installations it must be installed in the cluster or reachable from inside K8s.
- MongoDB or PostgreSQL are only required for the Tyk Self-managed installation and must be installed in the cluster, or reachable from inside K8s. If you are using the MongoDB or SQL pumps in the
tyk-ossinstallation you will require MongoDB or PostgreSQL installed for that as well. - For evaluation, you can install Redis with the
tyk-helm/simple-redischart. It is NOT recommended for production use. See Tyk Open Source installation for the Redis options.
For further detail on how to configure Tyk as an Ingress Gateway, or how to manage APIs in Tyk using the Kubernetes API, please refer to our Tyk Operator documentation. The Tyk Operator can be installed along this chart and works with all installation types.
The GitHub workflows are the source of truth for what gets tested:
unit-tests.yaml and
run-tests.yaml. The commands here reproduce those jobs
on your own machine. Where a local run cannot match CI, this section says so.
Run every snippet with bash. The loops and the unquoted --set braces below are bash
syntax, and they behave differently under other shells.
You need the following software. Install each from its own project, so the instructions match your platform.
| Software | Version | Why you need it |
|---|---|---|
| Docker | Any current release | kind runs the cluster inside a container. Docker Desktop and a standalone Docker Engine both work. |
| kubectl | Matching your cluster | Creating namespaces, reading secrets, and inspecting pods. |
| Helm | 3.x | Rendering and installing the charts. CI uses Helm 3. Helm 4 works, with one caveat below. |
| kind | Any current release | Creating the throwaway cluster that the smoke and integration tests run against. |
| Git | 2.17.0 or later | chart-testing reads git history to detect changed charts. |
You also need three test tools: chart-testing for linting, plus yamllint and yamale, which chart-testing calls but does not install.
Pick whichever route suits your platform:
# A package manager, if one carries all three
brew install chart-testing yamllint
# Or a chart-testing release binary plus pip for the two Python tools
# https://github.com/helm/chart-testing/releases
pip install yamllint yamaleMost Linux distributions also package yamllint. chart-testing only needs yamllint and
yamale on your PATH, so any install method works.
If you would rather not install the three on your host, the chart-testing container image bundles them.
Finally, install the unit-test plugin. This command is the same on every platform:
helm plugin install https://github.com/helm-unittest/helm-unittest.git --version v0.7.2 --verify=falseTwo notes on that command:
--verify=falseis required on Helm 4, which verifies plugin provenance and refuses a git source without it. Helm 3 accepts the command without the flag, which is why CI omits it.- The plugin carries its own Helm 3 SDK, so unit tests render through Helm 3 whichever Helm you install. A passing unit-test run tells you nothing about Helm 4 rendering.
Unit tests need no cluster and finish in seconds. Run them first when you change a template.
for chart in tyk-control-plane tyk-data-plane tyk-oss tyk-stack; do
helm dependency update "./$chart"
done
helm unittest \
./components/tyk-bootstrap \
./components/tyk-dashboard \
./components/tyk-dev-portal \
./components/tyk-pump \
./tyk-control-plane \
./tyk-stackThe umbrella charts pull their components in over file:// dependencies, so their suites
cannot render until helm dependency update vendors those into charts/.
CI discovers the chart list rather than hardcoding it, so a new suite is picked up without a
workflow edit. A chart qualifies only if it holds tests/*_test.yaml at its root. Charts
whose only tests/ directory sits under templates/ hold helm test hook pods instead, and
CI skips them.
helm repo add tyk-helm https://helm.tyk.io/public/helm/charts/
ct lint --config ct.yaml --allct lint covers the four umbrella charts only, because chart-dirs in ct.yaml
searches one level down. This command does not lint the charts under components/.
Linting also vendors the charts/ directories, so a later helm dependency update has
nothing left to do.
These need a cluster:
kind create cluster --name tyk-charts --config scripts/kind-cluster.yaml --wait 180sDelete it when you finish:
kind delete cluster --name tyk-chartsFor the install and helm test commands, follow the steps in
run-tests.yaml rather than a second copy kept here. The
workflow is what CI runs, and a duplicate drifts out of date.
Two of the three integration paths need credentials that this repository does not carry:
| Path | Requires |
|---|---|
tyk-oss |
Nothing beyond Redis |
tyk-stack |
A dashboard license, plus MongoDB |
tyk-data-plane |
HYBRID_RPC_KEY, HYBRID_API_KEY, and HYBRID_MDCB_HOST |
Export a license rather than writing it into a values file, so it stays out of any file you might commit:
export DASH_LICENSE='...'
helm install --namespace tyk-stack tyk-stack ./tyk-stack \
--set "global.license.dashboard=$DASH_LICENSE" ...The stack path also needs a MongoDB root password. You do not supply it: read it from the secret that the MongoDB chart creates, as the workflow does.
Read a local pass as weaker evidence than a CI pass, for four reasons:
- CI runs the smoke tests against five Kubernetes versions, v1.26.13 through v1.30.0. A kind
cluster with no pinned node image gives you whatever version kind currently defaults to,
which may be one CI never tests. Pass
--image kindest/node:v1.30.0to match. - CI pins chart-testing to v3.3.0 and uses Helm 3. Package managers generally carry newer versions of both.
- The
no-securitycontextjob asserts that no pod rendersrunAsUserorfsGroup. Linting renders that values file but makes no such assertion. - Both data-plane paths stay unrun without the
HYBRID_*credentials.
An install produces no output at all. If helm install against an oci:// URL prints
nothing and creates no release, check your Docker credential helper before you suspect the
chart. Helm calls that helper before it pulls an OCI chart, and a wedged helper blocks Helm
with no error. Find which helper you use, then probe it:
grep credsStore ~/.docker/config.json
echo registry-1.docker.io | docker-credential-<helper> getIf the helper does not answer within a few seconds, restart Docker. The --timeout flag does
not help here, because it covers the Kubernetes wait and Helm never gets that far.
An install times out on a first run. The gateway image is large, and kind keeps its image
cache inside the node container. kind delete cluster discards that cache, so every fresh
cluster downloads the images again. Raise --timeout on a slow connection.