Skip to content

Latest commit

 

History

History
626 lines (497 loc) · 34.3 KB

additional-configs.md

File metadata and controls

626 lines (497 loc) · 34.3 KB
Raw

This document describes additional options to configure your Tekton Pipelines installation.

Table of Contents

Configuring built-in remote Task and Pipeline resolution

Four remote resolvers are currently provided as part of the Tekton Pipelines installation. By default, these remote resolvers are enabled. Each resolver can be disabled by setting the appropriate feature flag in the resolvers-feature-flags ConfigMap in the tekton-pipelines-resolvers namespace:

  1. The bundles resolver, disabled by setting the enable-bundles-resolver feature flag to false.
  2. The git resolver, disabled by setting the enable-git-resolver feature flag to false.
  3. The hub resolver, disabled by setting the enable-hub-resolver feature flag to false.
  4. The cluster resolver, disabled by setting the enable-cluster-resolver feature flag to false.

Configuring CloudEvents notifications

When configured so, Tekton can generate CloudEvents for TaskRun, PipelineRun and CustomRunlifecycle events. The main configuration parameter is the URL of the sink. When not set, no notification is generated.

apiVersion: v1
kind: ConfigMap
metadata:
  name: config-events
  namespace: tekton-pipelines
  labels:
    app.kubernetes.io/instance: default
    app.kubernetes.io/part-of: tekton-pipelines
data:
  formats: tektonv1
  sink: https://my-sink-url

The sink used to be configured in the config-defaults config map. This option is still available, but deprecated, and will be removed.

apiVersion: v1
kind: ConfigMap
metadata:
  name: config-defaults
  namespace: tekton-pipelines
  labels:
    app.kubernetes.io/instance: default
    app.kubernetes.io/part-of: tekton-pipelines
data:
  default-cloud-events-sink: https://my-sink-url

Additionally, CloudEvents for CustomRuns require an extra configuration to be enabled. This setting exists to avoid collisions with CloudEvents that might be sent by custom task controllers:

apiVersion: v1
kind: ConfigMap
metadata:
  name: feature-flags
  namespace: tekton-pipelines
  labels:
    app.kubernetes.io/instance: default
    app.kubernetes.io/part-of: tekton-pipelines
data:
  send-cloudevents-for-runs: true

Configuring self-signed cert for private registry

The SSL_CERT_DIR is set to /etc/ssl/certs as the default cert directory. If you are using a self-signed cert for private registry and the cert file is not under the default cert directory, configure your registry cert in the config-registry-cert ConfigMap with the key cert.

Configuring environment variables

Environment variables can be configured in the following ways, mentioned in order of precedence from lowest to highest.

  1. Implicit environment variables
  2. Step/StepTemplate environment variables
  3. Environment variables specified via a default PodTemplate.
  4. Environment variables specified via a PodTemplate.

The environment variables specified by a PodTemplate supercedes all other ways of specifying environment variables. However, there exists a configuration i.e. default-forbidden-env, the environment variable specified in this list cannot be updated via a PodTemplate.

For example:

apiVersion: v1
kind: ConfigMap
metadata:
  name: config-defaults
  namespace: tekton-pipelines
data:
  default-timeout-minutes: "50"
  default-service-account: "tekton"
  default-forbidden-env: "TEST_TEKTON"
---
apiVersion: tekton.dev/v1beta1
kind: Task
metadata:
  name: mytask
  namespace: default
spec:
  steps:
    - name: echo-env
      image: ubuntu
      command: ["bash", "-c"]
      args: ["echo $TEST_TEKTON "]
      env:
          - name: "TEST_TEKTON"
            value: "true"
---
apiVersion: tekton.dev/v1beta1
kind: TaskRun
metadata:
  name: mytaskrun
  namespace: default
spec:
  taskRef:
    name: mytask
  podTemplate:
    env:
        - name: "TEST_TEKTON"
          value: "false"

In the above example the environment variable TEST_TEKTON will not be overriden by value specified in podTemplate, because the config-default option default-forbidden-env is configured with value TEST_TEKTON.

Customizing basic execution parameters

You can specify your own values that replace the default service account (ServiceAccount), timeout (Timeout), resolver (Resolver), and Pod template (PodTemplate) values used by Tekton Pipelines in TaskRun and PipelineRun definitions. To do so, modify the ConfigMap config-defaults with your desired values.

The example below customizes the following:

  • the default service account from default to tekton.
  • the default timeout from 60 minutes to 20 minutes.
  • the default app.kubernetes.io/managed-by label is applied to all Pods created to execute TaskRuns.
  • the default Pod template to include a node selector to select the node where the Pod will be scheduled by default. A list of supported fields is available here. For more information, see PodTemplate in TaskRuns or PodTemplate in PipelineRuns.
  • the default Workspace configuration can be set for any Workspaces that a Task declares but that a TaskRun does not explicitly provide.
  • the default maximum combinations of Parameters in a Matrix that can be used to fan out a PipelineTask. For more information, see Matrix.
  • the default resolver type to git.
apiVersion: v1
kind: ConfigMap
metadata:
  name: config-defaults
data:
  default-service-account: "tekton"
  default-timeout-minutes: "20"
  default-pod-template: |
    nodeSelector:
      kops.k8s.io/instancegroup: build-instance-group
  default-managed-by-label-value: "my-tekton-installation"
  default-task-run-workspace-binding: |
    emptyDir: {}
  default-max-matrix-combinations-count: "1024"
  default-resolver-type: "git"

Note: The _example key in the provided config-defaults.yaml file lists the keys you can customize along with their default values.

Customizing the Pipelines Controller behavior

To customize the behavior of the Pipelines Controller, modify the ConfigMap feature-flags via kubectl edit configmap feature-flags -n tekton-pipelines.

Note: Changing feature flags may result in undefined behavior for TaskRuns and PipelineRuns that are running while the change occurs.

The flags in this ConfigMap are as follows:

  • disable-affinity-assistant - set this flag to true to disable the Affinity Assistant that is used to provide Node Affinity for TaskRun pods that share workspace volume. The Affinity Assistant is incompatible with other affinity rules configured for TaskRun pods.

    Note: This feature flag is deprecated and will be removed in release v0.60. Consider using coschedule feature flag to configure Affinity Assistant behavior.

    Note: Affinity Assistant use Inter-pod affinity and anti-affinity that require substantial amount of processing which can slow down scheduling in large clusters significantly. We do not recommend using them in clusters larger than several hundred nodes

    Note: Pod anti-affinity requires nodes to be consistently labelled, in other words every node in the cluster must have an appropriate label matching topologyKey. If some or all nodes are missing the specified topologyKey label, it can lead to unintended behavior.

  • coschedule: set this flag determines how PipelineRun Pods are scheduled with Affinity Assistant. Acceptable values are "workspaces" (default), "pipelineruns", "isolate-pipelinerun", or "disabled". Setting it to "workspaces" will schedule all the taskruns sharing the same PVC-based workspace in a pipelinerun to the same node. Setting it to "pipelineruns" will schedule all the taskruns in a pipelinerun to the same node. Setting it to "isolate-pipelinerun" will schedule all the taskruns in a pipelinerun to the same node, and only allows one pipelinerun to run on a node at a time. Setting it to "disabled" will not apply any coschedule policy.

  • await-sidecar-readiness: set this flag to "false" to allow the Tekton controller to start a TasksRun's first step immediately without waiting for sidecar containers to be running first. Using this option should decrease the time it takes for a TaskRun to start running, and will allow TaskRun pods to be scheduled in environments that don't support Downward API volumes (e.g. some virtual kubelet implementations). However, this may lead to unexpected behaviour with Tasks that use sidecars, or in clusters that use injected sidecars (e.g. Istio). Setting this flag to "false" will mean the running-in-environment-with-injected-sidecars flag has no effect.

  • running-in-environment-with-injected-sidecars: set this flag to "false" to allow the Tekton controller to start a TasksRun's first step immediately if it has no Sidecars specified. Using this option should decrease the time it takes for a TaskRun to start running. However, for clusters that use injected sidecars (e.g. Istio) this can lead to unexpected behavior.

  • require-git-ssh-secret-known-hosts: set this flag to "true" to require that Git SSH Secrets include a known_hosts field. This ensures that a git remote server's key is validated before data is accepted from it when authenticating over SSH. Secrets that don't include a known_hosts will result in the TaskRun failing validation and not running.

  • enable-tekton-oci-bundles: set this flag to "true" to enable the tekton OCI bundle usage (see the tekton bundle contract). Enabling this option allows the use of bundle field in taskRef and pipelineRef for Pipeline, PipelineRun and TaskRun. By default, this option is disabled ("false"), which means it is disallowed to use the bundle field.

  • disable-creds-init - set this flag to "true" to disable Tekton's built-in credential initialization and use Workspaces to mount credentials from Secrets instead. The default is false. For more information, see the associated issue.

  • enable-api-fields: When using v1beta1 APIs, setting this field to "stable" or "beta" enables beta features. When using v1 APIs, setting this field to "stable" allows only stable features, and setting it to "beta" allows only beta features. Set this field to "alpha" to allow alpha features to be used.

For example:

apiVersion: v1
kind: ConfigMap
metadata:
  name: feature-flags
data:
  enable-api-fields: "alpha" # Allow alpha fields to be used in Tasks and Pipelines.

  • trusted-resources-verification-no-match-policy: Setting this flag to fail will fail the taskrun/pipelinerun if no matching policies found. Setting to warn will skip verification and log a warning if no matching policies are found, but not fail the taskrun/pipelinerun. Setting to ignore will skip verification if no matching policies found. Defaults to "ignore".

  • results-from: set this flag to "termination-message" to use the container's termination message to fetch results from. This is the default method of extracting results. Set it to "sidecar-logs" to enable use of a results sidecar logs to extract results instead of termination message.

  • enable-provenance-in-status: Set this flag to "true" to enable populating the provenance field in TaskRun and PipelineRun status. The provenance field contains metadata about resources used in the TaskRun/PipelineRun such as the source from where a remote Task/Pipeline definition was fetched. By default, this is set to true. To disable populating this field, set this flag to "false".

  • set-security-context: Set this flag to true to set a security context for containers injected by Tekton that will allow TaskRun pods to run in namespaces with restricted pod security admission. By default, this is set to false.

Alpha Features

Alpha features in the following table are still in development and their syntax is subject to change.

  • To enable the features without an individual flag: set the enable-api-fields feature flag to "alpha" in the feature-flags ConfigMap alongside your Tekton Pipelines deployment via kubectl patch cm feature-flags -n tekton-pipelines -p '{"data":{"enable-api-fields":"alpha"}}'.
  • To enable the features with an individual flag: set the individual flag accordingly in the feature-flag ConfigMap alongside your Tekton Pipelines deployment. Example: kubectl patch cm feature-flags -n tekton-pipelines -p '{"data":{"<FLAG-NAME>":"<FLAG-VALUE>"}}'.

Features currently in "alpha" are:

Feature Proposal Release Individual Flag
Bundles TEP-0005 v0.18.0 enable-tekton-oci-bundles
Hermetic Execution Mode TEP-0025 v0.25.0
Windows Scripts TEP-0057 v0.28.0
Debug TEP-0042 v0.26.0
Step and Sidecar Overrides TEP-0094 v0.34.0
Matrix TEP-0090 v0.38.0
Task-level Resource Requirements TEP-0104 v0.39.0
Trusted Resources TEP-0091 N/A trusted-resources-verification-no-match-policy
Larger Results via Sidecar Logs TEP-0127 v0.43.0 results-from
Configure Default Resolver TEP-0133 N/A
Coschedule TEP-0135 N/A coschedule

Beta Features

Beta features are fields of stable CRDs that follow our "beta" compatibility policy. To enable these features, set the enable-api-fields feature flag to "beta" in the feature-flags ConfigMap alongside your Tekton Pipelines deployment via kubectl patch cm feature-flags -n tekton-pipelines -p '{"data":{"enable-api-fields":"beta"}}'.

For beta versions of Tekton CRDs, setting enable-api-fields to "beta" is the same as setting it to "stable", except where otherwise noted.

Features currently in "beta" are:

Feature Proposal Alpha Release Beta Release Individual Flag enable-api-fields=beta required for v1beta1
Array Results and Array Indexing TEP-0076 v0.38.0 v0.45.0 No
Object Parameters and Results TEP-0075 v0.46.0 No
Remote Tasks and Remote Pipelines TEP-0060 v0.41.0 No
Provenance field in Status issue#5550 v0.41.0 v0.48.0 enable-provenance-in-status No
Isolated Step & Sidecar Workspaces TEP-0029 v0.24.0 v0.50.0 Yes

Enabling larger results using sidecar logs

Note: The maximum size of a Task's results is limited by the container termination message feature of Kubernetes, as results are passed back to the controller via this mechanism. At present, the limit is per task is “4096 bytes”. All results produced by the task share this upper limit.

To exceed this limit of 4096 bytes, you can enable larger results using sidecar logs. By enabling this feature, you will have a configurable limit (with a default of 4096 bytes) per result with no restriction on the number of results. The results are still stored in the taskRun CRD, so they should not exceed the 1.5MB CRD size limit.

Note: to enable this feature, you need to grant get access to all pods/log to the tekton-pipelines-controller. This means that the tekton pipeline controller has the ability to access the pod logs.

  1. Create a cluster role and rolebinding by applying the following spec to provide log access to tekton-pipelines-controller.
kubectl apply -f optional_config/enable-log-access-to-controller/
  1. Set the results-from feature flag to use sidecar logs by setting results-from: sidecar-logs in the configMap.
kubectl patch cm feature-flags -n tekton-pipelines -p '{"data":{"results-from":"sidecar-logs"}}'
  1. If you want the size per result to be something other than 4096 bytes, you can set the max-result-size feature flag in bytes by setting max-result-size: 8192(whatever you need here). Note: The value you can set here cannot exceed the size of the CRD limit of 1.5 MB.
kubectl patch cm feature-flags -n tekton-pipelines -p '{"data":{"max-result-size":"<VALUE-IN-BYTES>"}}'

Configuring High Availability

If you want to run Tekton Pipelines in a way so that webhooks are resiliant against failures and support high concurrency scenarios, you need to run a Metrics Server in your Kubernetes cluster. This is required by the Horizontal Pod Autoscalers to compute replica count.

See HA Support for Tekton Pipeline Controllers for instructions on configuring High Availability in the Tekton Pipelines Controller.

The default configuration is defined in webhook-hpa.yaml which can be customized to better fit specific usecases.

Configuring tekton pipeline controller performance

Out-of-the-box, Tekton Pipelines Controller is configured for relatively small-scale deployments but there have several options for configuring Pipelines' performance are available. See the Performance Configuration document which describes how to change the default ThreadsPerController, QPS and Burst settings to meet your requirements.

Running TaskRuns and PipelineRuns with restricted pod security standards

To allow TaskRuns and PipelineRuns to run in namespaces with restricted pod security standards, set the "set-security-context" feature flag to "true" in the feature-flags configMap. This configuration option applies a SecurityContext to any containers injected into TaskRuns by the Pipelines controller. This SecurityContext may not be supported in all Kubernetes implementations (for example, OpenShift).

Note: running TaskRuns and PipelineRuns in the "tekton-pipelines" namespace is discouraged.

Platform Support

The Tekton project provides support for running on x86 Linux Kubernetes nodes.

The project produces images capable of running on other architectures and operating systems, but may not be able to help debug issues specific to those platforms as readily as those that affect Linux on x86.

The controller and webhook components are currently built for:

  • linux/amd64
  • linux/arm64
  • linux/arm (Arm v7)
  • linux/ppc64le (PowerPC)
  • linux/s390x (IBM Z)

The entrypoint component is also built for Windows, which enables TaskRun workloads to execute on Windows nodes. See Windows documentation for more information.

Creating a custom release of Tekton Pipelines

You can create a custom release of Tekton Pipelines by following and customizing the steps in Creating an official release. For example, you might want to customize the container images built and used by Tekton Pipelines.

Verify Tekton Pipelines Release

We will refine this process over time to be more streamlined. For now, please follow the steps listed in this section to verify Tekton pipeline release.

Tekton Pipeline's images are being signed by Tekton Chains since 0.27.1. You can verify the images with cosign using the Tekton's public key.

Verify signatures using cosign

With Go 1.16+, you can install cosign by running:

go install github.com/sigstore/cosign/cmd/cosign@latest

You can verify Tekton Pipelines official images using the Tekton public key:

cosign verify -key https://raw.githubusercontent.com/tektoncd/chains/main/tekton.pub gcr.io/tekton-releases/github.com/tektoncd/pipeline/cmd/controller:v0.28.1

which results in:

Verification for gcr.io/tekton-releases/github.com/tektoncd/pipeline/cmd/controller:v0.28.1 --
The following checks were performed on each of these signatures:
  - The cosign claims were validated
  - The signatures were verified against the specified public key
  - Any certificates were verified against the Fulcio roots.
{
  "Critical": {
    "Identity": {
      "docker-reference": "gcr.io/tekton-releases/github.com/tektoncd/pipeline/cmd/controller"
    },
    "Image": {
      "Docker-manifest-digest": "sha256:0c320bc09e91e22ce7f01e47c9f3cb3449749a5f72d5eaecb96e710d999c28e8"
    },
    "Type": "Tekton container signature"
  },
  "Optional": {}
}

The verification shows a list of checks performed and returns the digest in Critical.Image.Docker-manifest-digest which can be used to retrieve the provenance from the transparency logs for that image using rekor-cli.

Verify the transparency logs using rekor-cli

Install the rekor-cli by running:

go install -v github.com/sigstore/rekor/cmd/rekor-cli@latest

Now, use the digest collected from the previous section in Critical.Image.Docker-manifest-digest, for example, sha256:0c320bc09e91e22ce7f01e47c9f3cb3449749a5f72d5eaecb96e710d999c28e8.

Search the transparency log with the digest just collected:

rekor-cli search --sha sha256:0c320bc09e91e22ce7f01e47c9f3cb3449749a5f72d5eaecb96e710d999c28e8

which results in:

Found matching entries (listed by UUID):
68a53d0e75463d805dc9437dda5815171502475dd704459a5ce3078edba96226

Tekton Chains generates provenance based on the custom format in which the subject holds the list of artifacts which were built as part of the release. For the Pipeline release, subject includes a list of images including pipeline controller, pipeline webhook, etc. Use the UUID to get the provenance:

rekor-cli get --uuid 68a53d0e75463d805dc9437dda5815171502475dd704459a5ce3078edba96226 --format json | jq -r .Attestation | base64 --decode | jq

which results in:

{
  "_type": "https://in-toto.io/Statement/v0.1",
  "predicateType": "https://tekton.dev/chains/provenance",
  "subject": [
    {
      "name": "gcr.io/tekton-releases/github.com/tektoncd/pipeline/cmd/controller",
      "digest": {
        "sha256": "0c320bc09e91e22ce7f01e47c9f3cb3449749a5f72d5eaecb96e710d999c28e8"
      }
    },
    {
      "name": "gcr.io/tekton-releases/github.com/tektoncd/pipeline/cmd/entrypoint",
      "digest": {
        "sha256": "2fa7f7c3408f52ff21b2d8c4271374dac4f5b113b1c4dbc7d5189131e71ce721"
      }
    },
    {
      "name": "gcr.io/tekton-releases/github.com/tektoncd/pipeline/cmd/git-init",
      "digest": {
        "sha256": "83d5ec6addece4aac79898c9631ee669f5fee5a710a2ed1f98a6d40c19fb88f7"
      }
    },
    {
      "name": "gcr.io/tekton-releases/github.com/tektoncd/pipeline/cmd/imagedigestexporter",
      "digest": {
        "sha256": "e4d77b5b8902270f37812f85feb70d57d6d0e1fed2f3b46f86baf534f19cd9c0"
      }
    },
    {
      "name": "gcr.io/tekton-releases/github.com/tektoncd/pipeline/cmd/nop",
      "digest": {
        "sha256": "59b5304bcfdd9834150a2701720cf66e3ebe6d6e4d361ae1612d9430089591f8"
      }
    },
    {
      "name": "gcr.io/tekton-releases/github.com/tektoncd/pipeline/cmd/pullrequest-init",
      "digest": {
        "sha256": "4992491b2714a73c0a84553030e6056e6495b3d9d5cc6b20cf7bc8c51be779bb"
      }
    },
    {
      "name": "gcr.io/tekton-releases/github.com/tektoncd/pipeline/cmd/webhook",
      "digest": {
        "sha256": "bf0ef565b301a1981cb2e0d11eb6961c694f6d2401928dccebe7d1e9d8c914de"
      }
    }
  ],
  ...

Now, verify the digest in the release.yaml by matching it with the provenance, for example, the digest for the release v0.28.1:

curl -s https://storage.googleapis.com/tekton-releases/pipeline/previous/v0.28.1/release.yaml | grep github.com/tektoncd/pipeline/cmd/controller:v0.28.1 | awk -F"github.com/tektoncd/pipeline/cmd/controller:v0.28.1@" '{print $2}'

which results in:

sha256:0c320bc09e91e22ce7f01e47c9f3cb3449749a5f72d5eaecb96e710d999c28e8

Now, you can verify the deployment specifications in the release.yaml to match each of these images and their digest. The tekton-pipelines-controller deployment specification has a container named tekton-pipeline-controller and a list of image references with their digest as part of the args:

      containers:
        - name: tekton-pipelines-controller
          image: gcr.io/tekton-releases/github.com/tektoncd/pipeline/cmd/controller:v0.28.1@sha256:0c320bc09e91e22ce7f01e47c9f3cb3449749a5f72d5eaecb96e710d999c28e8
          args: [
            # These images are built on-demand by `ko resolve` and are replaced
            # by image references by digest.
              "-git-image",
              "gcr.io/tekton-releases/github.com/tektoncd/pipeline/cmd/git-init:v0.28.1@sha256:83d5ec6addece4aac79898c9631ee669f5fee5a710a2ed1f98a6d40c19fb88f7",
              "-entrypoint-image",
              "gcr.io/tekton-releases/github.com/tektoncd/pipeline/cmd/entrypoint:v0.28.1@sha256:2fa7f7c3408f52ff21b2d8c4271374dac4f5b113b1c4dbc7d5189131e71ce721",
              "-nop-image",
              "gcr.io/tekton-releases/github.com/tektoncd/pipeline/cmd/nop:v0.28.1@sha256:59b5304bcfdd9834150a2701720cf66e3ebe6d6e4d361ae1612d9430089591f8",
              "-imagedigest-exporter-image",
              "gcr.io/tekton-releases/github.com/tektoncd/pipeline/cmd/imagedigestexporter:v0.28.1@sha256:e4d77b5b8902270f37812f85feb70d57d6d0e1fed2f3b46f86baf534f19cd9c0",
              "-pr-image",
              "gcr.io/tekton-releases/github.com/tektoncd/pipeline/cmd/pullrequest-init:v0.28.1@sha256:4992491b2714a73c0a84553030e6056e6495b3d9d5cc6b20cf7bc8c51be779bb",

Similarly, you can verify the rest of the images which were published as part of the Tekton Pipelines release:

gcr.io/tekton-releases/github.com/tektoncd/pipeline/cmd/git-init
gcr.io/tekton-releases/github.com/tektoncd/pipeline/cmd/entrypoint
gcr.io/tekton-releases/github.com/tektoncd/pipeline/cmd/nop
gcr.io/tekton-releases/github.com/tektoncd/pipeline/cmd/imagedigestexporter
gcr.io/tekton-releases/github.com/tektoncd/pipeline/cmd/pullrequest-init
gcr.io/tekton-releases/github.com/tektoncd/pipeline/cmd/webhook

Verify Tekton Resources

Trusted Resources is a feature to verify Tekton Tasks and Pipelines. The current version of feature supports v1beta1 Task and Pipeline. For more details please take a look at Trusted Resources.

Pipelineruns with Affinity Assistant

The cluster operators can review the guidelines to cordon a node in the cluster with the tekton controller and the affinity assistant is enabled.

Next steps

To get started with Tekton check the [Introductory tutorials][quickstarts], the [how-to guides][howtos], and the [examples folder][examples].

{{% comment %}} Except as otherwise noted, the content of this page is licensed under the Creative Commons Attribution 4.0 License, and code samples are licensed under the Apache 2.0 License. {{% /comment %}}