Adding kind in place of minikube #147
Conversation
* The networking in minikube seems to be causing me problems locally I can only assume this is happening in CI Signed-off-by: Shawn Hurley <[email protected]>
WalkthroughThis PR introduces a new GitHub Actions workflow for Konveyor that orchestrates end-to-end API and UI integration tests. The workflow accepts configurable inputs for operator tags, component images, and test references, provisions a Kubernetes environment (kind cluster or Minikube), installs the tackle2 operator, and executes tests with conditional logic supporting multiple operator versions. Changes
Sequence Diagram(s)sequenceDiagram
participant Trigger as Workflow<br/>(Input Parameters)
participant API as e2e-api-<br/>integration-tests
participant UI as e2e-ui-<br/>integration-tests
participant Env as Environment<br/>(kind/Minikube)
participant Operator as Tackle2<br/>Operator
participant Tests as Tests<br/>(Go/Cypress)
Trigger->>API: Extract PR ref from description
Trigger->>UI: Extract PR ref from description
API->>Env: Setup kind cluster
UI->>Env: Setup Minikube
Note over Env: Conditional based on operator_tag<br/>(v0.7 / v0.8 / latest)
API->>Operator: Install operator version
UI->>Operator: Install operator + Java provider
API->>Tests: Run Go API tests (tiers)
UI->>Tests: Run Cypress UI tests
Tests-->>Trigger: Generate test summaries & artifacts
Estimated code review effort🎯 4 (Complex) | ⏱️ ~45 minutes
Possibly related PRs
Suggested reviewers
Poem
Pre-merge checks and finishing touches❌ Failed checks (1 warning)
✅ Passed checks (2 passed)
✨ Finishing touches🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 9
🧹 Nitpick comments (1)
.github/workflows/kind-global-ci.yml (1)
315-315: Simplify pull request extraction regex patterns.Lines 182 and 315 use case-insensitive character classes
[A|a],[P|p], etc., which are unnecessary and harder to read. Use theiflag for case-insensitive matching instead:- PULL_REQUEST_NUMBER=$(echo ${body} | grep -oP '[A|a][P|p][I|i] [T|t]ests [P|p][R|r]:\s*\K\d+' || true) + PULL_REQUEST_NUMBER=$(echo ${body} | grep -oiP 'api tests pr:\s*\K\d+' || true)Similarly for the UI extraction on line 315.
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (1)
.github/workflows/kind-global-ci.yml(1 hunks)
🧰 Additional context used
🪛 actionlint (1.7.8)
.github/workflows/kind-global-ci.yml
266-266: the runner of "actions/setup-go@v4" action is too old to run on GitHub Actions. update the action's version to fix this issue
(action)
333-333: the runner of "docker/setup-buildx-action@v2" action is too old to run on GitHub Actions. update the action's version to fix this issue
(action)
459-459: property "test-tags" is not defined in object type {api_hub_tests_ref: string; api_tests_ref: string; api_tests_tiers: string; component_name: string; operator_tag: string; run_api_tests: bool; run_ui_tests: bool; tag: string; ui_test_tags: string; ui_tests_ref: string}
(expression)
⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (4)
- GitHub Check: Verify api and ui tests run via
global-ci-bundle.ymlwith default input / e2e-api-integration-tests - GitHub Check: Verify api and ui tests run via
global-ci-bundle.ymlwith default input / e2e-ui-integration-tests - GitHub Check: Verify api and ui tests run via
global-ci.ymlwith default inputs / e2e-api-integration-tests - GitHub Check: Verify api and ui tests run via
global-ci.ymlwith default inputs / e2e-ui-integration-tests
🔇 Additional comments (2)
.github/workflows/kind-global-ci.yml (2)
296-297: Clarify scope mismatch: UI tests still use Minikube, not kind.The PR objective states "Adding kind in place of minikube," but the e2e-ui-integration-tests job still uses Minikube (line 336–352) rather than kind. Is this intentional (e.g., UI tests require Minikube-specific features), or should UI tests migrate to kind as well?
Please clarify whether UI tests should also migrate to kind or if using Minikube is a deliberate choice. This affects the broader validation of the PR's goal.
349-349: Fix nonsensical string matching for latest operator tag.Line 349 checks
startsWith(inputs.operator_tag, 'latest'), which will never be true for the string"latest". Compare line 255, which correctly uses equality:inputs.operator_tag == 'latest'.- if: "${{ startsWith(inputs.operator_tag, 'latest') }}" + if: "${{ inputs.operator_tag == 'latest' }}"Likely an incorrect or invalid review comment.
| - uses: actions/setup-go@v4 | ||
| with: | ||
| go-version: 1.21 |
There was a problem hiding this comment.
🧩 Analysis chain
Update outdated GitHub Actions.
Line 266: actions/setup-go@v4 is outdated per static analysis. Update to v5 or the latest stable version.
🌐 Web query:
What is the latest version of actions/setup-go?
💡 Result:
The latest version is actions/setup-go v6.0.0 (current as of Oct 31, 2025). [1][2]
Sources:
[1] actions/setup-go releases on GitHub.
[2] actions/setup-go package page (latest: 6.0.0).
Update actions/setup-go to the latest version.
Line 266: actions/setup-go@v4 should be updated to v6.0.0, the latest stable version.
- uses: actions/setup-go@v6
with:
go-version: 1.21
🧰 Tools
🪛 actionlint (1.7.8)
266-266: the runner of "actions/setup-go@v4" action is too old to run on GitHub Actions. update the action's version to fix this issue
(action)
🤖 Prompt for AI Agents
In .github/workflows/kind-global-ci.yml around lines 266 to 268, the workflow
uses actions/setup-go@v4 which should be updated to the latest stable v6
release; change the action reference to actions/setup-go@v6 (or
actions/[email protected]) so the workflow uses the current major version while
keeping the existing go-version: 1.21 configuration.
|
|
||
| - name: Build and run golang API tests | ||
| run: | | ||
| export HUB_BASE_URL="http://$(minikube ip)/hub" |
There was a problem hiding this comment.
Fix hardcoded minikube IP command in kind-based API test job.
Line 277 exports HUB_BASE_URL using minikube ip, but the e2e-api-integration-tests job runs on a kind cluster (line 201), not Minikube. This will fail at runtime. Use kind-appropriate commands to retrieve the cluster IP or service endpoint.
For kind clusters, retrieve the control plane IP using:
export HUB_BASE_URL="http://$(kubectl get node -o jsonpath='{.items[0].status.addresses[?(@.type=="InternalIP")].address}'):8080/hub"Alternatively, if the Hub service is exposed via a LoadBalancer or port-forward, adjust accordingly.
🤖 Prompt for AI Agents
.github/workflows/kind-global-ci.yml around line 277: the workflow exports
HUB_BASE_URL using minikube ip which is wrong for the kind-based
e2e-api-integration-tests job; replace the minikube-specific command with a
kind-appropriate approach — obtain the control-plane node internal IP via
kubectl (jsonpath to
.items[0].status.addresses[?(@.type=="InternalIP")].address) and build the URL
with the Hub port and /hub path, or alternatively set HUB_BASE_URL to the
service endpoint if the Hub is exposed via LoadBalancer or use kubectl
port-forward to expose the Hub and point HUB_BASE_URL to the forwarded localhost
URL.
| # TODO Should DRY this | ||
| - name: set up docker buildx | ||
| if: "${{ inputs.component_name != '' }}" | ||
| uses: docker/setup-buildx-action@v2 |
There was a problem hiding this comment.
🧩 Analysis chain
Update outdated GitHub Actions.
Line 333: docker/setup-buildx-action@v2 is outdated per static analysis. Update to v3 or the latest stable version.
🌐 Web query:
What is the latest version of docker/setup-buildx-action?
💡 Result:
The latest release is docker/setup-buildx-action v3.11.1 (released June 18, 2025). [1][2]
Sources:
[1] GitHub releases — docker/setup-buildx-action.
[2] GitHub repo README / Marketplace entry for docker/setup-buildx-action.
🏁 Script executed:
#!/bin/bash
# Check the actual version of docker/setup-buildx-action in the workflow file
sed -n '330,336p' .github/workflows/kind-global-ci.ymlLength of output: 269
Update outdated GitHub Actions.
Line 333: docker/setup-buildx-action@v2 is outdated. Update to v3 or the latest stable version (v3.11.1, released June 18, 2025).
🧰 Tools
🪛 actionlint (1.7.8)
333-333: the runner of "docker/setup-buildx-action@v2" action is too old to run on GitHub Actions. update the action's version to fix this issue
(action)
🤖 Prompt for AI Agents
In .github/workflows/kind-global-ci.yml around line 333, the workflow references
docker/setup-buildx-action@v2 which is outdated; update the action reference to
the latest stable version (for example docker/setup-buildx-action@v3 or
docker/[email protected]). Edit the workflow line to replace the v2
tag with the chosen v3 tag, run a quick workflow syntax/lint check, and push the
change so CI uses the updated action.
| - name: start minikube | ||
| uses: konveyor/tackle2-operator/.github/actions/[email protected] | ||
| if: "${{ startsWith(inputs.operator_tag, 'v0.7') }}" | ||
| with: | ||
| memory: 'max' | ||
| cpus: 'max' | ||
| - name: start minikube | ||
| uses: konveyor/tackle2-operator/.github/actions/[email protected] | ||
| if: "${{ startsWith(inputs.operator_tag, 'v0.8') }}" | ||
| with: | ||
| memory: 'max' | ||
| cpus: 'max' | ||
| - name: start minikube | ||
| uses: konveyor/tackle2-operator/.github/actions/start-minikube@main | ||
| if: "${{ startsWith(inputs.operator_tag, 'latest') }}" | ||
| with: | ||
| memory: 'max' | ||
| cpus: 'max' |
There was a problem hiding this comment.
🛠️ Refactor suggestion | 🟠 Major
Extract repeated Minikube startup logic into a reusable action or step.
Lines 335–352 contain three nearly identical start-minikube steps conditioned on operator version (v0.7, v0.8, latest). This violates DRY and increases maintenance risk. Use a matrix strategy, a composite action, or a single conditional with version-parameterized inputs.
🤖 Prompt for AI Agents
.github/workflows/kind-global-ci.yml around lines 335 to 352: the three
start-minikube steps are duplicated with only the action ref and if-condition
differing; replace them with a single reusable approach—either use a matrix over
the three operator_tag cases, create a small composite action that accepts the
operator tag and delegates to the appropriate ref, or consolidate into one step
that computes the correct action ref from inputs.operator_tag and passes memory
and cpus once—ensure the chosen solution preserves the conditional behavior for
v0.7, v0.8 and latest and keeps memory:'max' and cpus:'max'.
| - name: install konveyor | ||
| uses: konveyor/tackle2-operator/.github/actions/[email protected] | ||
| if: "${{ startsWith(inputs.operator_tag, 'v0.7') }}" | ||
| with: | ||
| operator-bundle-image: "quay.io/konveyor/tackle2-operator-bundle:${{ inputs.operator_tag }}" | ||
| hub-image: "quay.io/konveyor/tackle2-hub:${{ inputs.tag }}" | ||
| ui-image: "quay.io/konveyor/tackle2-ui:${{ inputs.tag }}" | ||
| addon-analyzer-image: "quay.io/konveyor/tackle2-addon-analyzer:${{ inputs.tag }}" | ||
| java-provider-image: "quay.io/konveyor/java-external-provider:${{ inputs.tag }}" | ||
| image-pull-policy: IfNotPresent | ||
| analyzer-container-memory: 0 | ||
| analyzer-container-cpu: 0 | ||
| - name: install konveyor | ||
| uses: konveyor/tackle2-operator/.github/actions/[email protected] | ||
| if: "${{ startsWith(inputs.operator_tag, 'v0.8') }}" | ||
| with: | ||
| operator-bundle-image: "quay.io/konveyor/tackle2-operator-bundle:${{ inputs.operator_tag }}" | ||
| hub-image: "quay.io/konveyor/tackle2-hub:${{ inputs.tag }}" | ||
| ui-image: "quay.io/konveyor/tackle2-ui:${{ inputs.tag }}" | ||
| addon-analyzer-image: "quay.io/konveyor/tackle2-addon-analyzer:${{ inputs.tag }}" | ||
| java-provider-image: "quay.io/konveyor/java-external-provider:${{ inputs.tag }}" | ||
| image-pull-policy: IfNotPresent | ||
| analyzer-container-memory: 0 | ||
| analyzer-container-cpu: 0 | ||
| - name: install konveyor | ||
| uses: konveyor/tackle2-operator/.github/actions/install-tackle@main | ||
| if: "${{ inputs.operator_tag == 'latest'}}" | ||
| with: | ||
| operator-bundle-image: "quay.io/konveyor/tackle2-operator-bundle:${{ inputs.operator_tag }}" | ||
| hub-image: "quay.io/konveyor/tackle2-hub:${{ inputs.tag }}" | ||
| ui-image: "quay.io/konveyor/tackle2-ui:${{ inputs.tag }}" | ||
| addon-analyzer-image: "quay.io/konveyor/tackle2-addon-analyzer:${{ inputs.tag }}" | ||
| java-provider-image: "quay.io/konveyor/java-external-provider:${{ inputs.tag }}" | ||
| image-pull-policy: IfNotPresent | ||
| analyzer-container-memory: 0 | ||
| analyzer-container-cpu: 0 |
There was a problem hiding this comment.
🛠️ Refactor suggestion | 🟠 Major
Extract repeated operator installation logic into a reusable action or step.
Lines 375–410 contain three nearly identical install-tackle steps conditioned on operator version, mirroring the duplication in lines 232–263. Consolidate using the same refactoring strategy applied to the API tests job.
🤖 Prompt for AI Agents
.github/workflows/kind-global-ci.yml lines 375-410: three nearly identical
"install konveyor" steps conditional on operator_tag should be consolidated;
extract the common inputs into a single reusable step (either a composite action
in .github/actions or a workflow-level reusable job/step) and drive the operator
ref via a parameter or a conditional expression (use a single uses:
konveyor/tackle2-operator/.github/actions/install-tackle@<ref> where <ref> is
computed from inputs.operator_tag or pass operator ref as an input), then
replace the three blocks with one step that supplies operator-bundle-image,
hub-image, ui-image, addon-analyzer-image, java-provider-image,
image-pull-policy, analyzer-container-memory and analyzer-container-cpu from
variables/inputs to avoid duplication.
| uses: actions/upload-artifact@v4 | ||
| if: always() | ||
| with: | ||
| name: tackle-ui-test-reports-${{ inputs.test-tags }} |
There was a problem hiding this comment.
Fix undefined input property reference.
Line 459 references inputs.test-tags, but the input is named ui_test_tags (line 78). This will cause a runtime error.
- name: tackle-ui-test-reports-${{ inputs.test-tags }}
+ name: tackle-ui-test-reports-${{ inputs.ui_test_tags }}🧰 Tools
🪛 actionlint (1.7.8)
459-459: property "test-tags" is not defined in object type {api_hub_tests_ref: string; api_tests_ref: string; api_tests_tiers: string; component_name: string; operator_tag: string; run_api_tests: bool; run_ui_tests: bool; tag: string; ui_test_tags: string; ui_tests_ref: string}
(expression)
🤖 Prompt for AI Agents
In .github/workflows/kind-global-ci.yml around line 459, the step name uses an
undefined input reference `inputs.test-tags`; replace it with the correct input
`inputs.ui_test_tags` to match the input declared at line 78 so the workflow
resolves the value at runtime (update any other occurrences of
`inputs.test-tags` in this file to `inputs.ui_test_tags` as well).
1 participant
Summary by CodeRabbit