feat(Kiali): Kiali multicluster evals - #1312
Conversation
|
👋 Heads up — this pull request changes files owned by @aljesusg. You are listed as an owner of one or more of the changed areas in |
257ef81 to
8fb31ab
Compare
cajieh
left a comment
cajieh
left a comment
There was a problem hiding this comment.
Added a few comments, but I still need to do another pass.
Als, should these 6 commits be squashed, or is keeping them separate intentional?
| needs: check-trigger | ||
| if: | | ||
| needs.check-trigger.outputs.should-run == 'true' && | ||
| needs.check-trigger.outputs.label-selector == 'suite=kiali' |
There was a problem hiding this comment.
Curious about this gate — does check-trigger currently expose a label-selector output? On main I’m only seeing should-run, matrix, pr-number, pr-sha, and is-pr. If label-selector isn’t set, would this job always be skipped?
Also for weekly schedule runs, suite selection looks like a matrix (schedule-matrix) rather than a single suite=kiali — how is multicluster supposed to be triggered in that case?
There was a problem hiding this comment.
Good catch. That was written against the pre-matrix workflow. After rebasing onto main, check-trigger no longer exposes label-selector.
Updated to a run-multicluster output set when the selected suite is kiali, all, or schedule-matrix (weekly runs). The multicluster job now gates on run-multicluster == 'true', so it still runs on weekly schedules and on /run-mcpchecker kiali.
| always() && | ||
| github.event_name == 'schedule' && | ||
| needs.check-trigger.outputs.label-selector == 'suite=kiali' && | ||
| needs.run-evaluation-multicluster.result == 'success' |
There was a problem hiding this comment.
Same question as on run-evaluation-multicluster: if label-selector isn’t an output of check-trigger, would this scheduled commit job ever run?
| COMMIT_SHA: ${{ needs.check-trigger.outputs.pr-sha }} | ||
| RUN_URL: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }} | ||
| run: | | ||
| BRANCH="chore/update-eval-results" |
There was a problem hiding this comment.
Is it intentional to reuse chore/update-eval-results here? The existing results committer uses the same branch with git push -f — could weekly kiali + multicluster race and overwrite each other? Would a separate branch name be safer?
There was a problem hiding this comment.
Agreed on the race risk with two jobs force-pushing the same branch.
To avoid that, multicluster results are now copied inside the existing commit-results job (single branch / single force-push) instead of a second commit-multicluster-results job.
b1eb77f to
a99b27a
Compare
|
@cajieh thanks for the review! Related the commits, I'll squash on merge as usual. |
Signed-off-by: josunect <jcordoba@redhat.com> Signed-off-by: Josune Cordoba <jcordoba@redhat.com>
Make invokes /bin/sh by default; dash does not support pipefail used by setup-kiali-multicluster and related targets. Co-authored-by: Cursor <cursoragent@cursor.com> Signed-off-by: josunect <jcordoba@redhat.com> Signed-off-by: Josune Cordoba <jcordoba@redhat.com>
-kv dev pushes a local image but Kiali only builds it when KIALI_BUILD_DEV_IMAGE=true. Export that for dev setups and add Node 24 + corepack to the multicluster workflow job. Also fix check-trigger suite output wiring so the multicluster job runs when suite is kiali. Co-authored-by: Cursor <cursoragent@cursor.com> Signed-off-by: josunect <jcordoba@redhat.com> Signed-off-by: Josune Cordoba <jcordoba@redhat.com>
Signed-off-by: josunect <jcordoba@redhat.com> Signed-off-by: Josune Cordoba <jcordoba@redhat.com>
Signed-off-by: Josune Cordoba <jcordoba@redhat.com>
Signed-off-by: Josune Cordoba <jcordoba@redhat.com>
d987902 to
ca7cf67
Compare
2 participants
Summary
Adds multicluster Kiali mcpchecker evals against a primary-remote Kind setup (
easthome cluster,westremote cluster).evals/tasks/kiali/multicluster/covering mesh status, listing mesh clusters, remote metrics/workloads/topology, and traces (agent must identify the remote cluster; judge checks forwest).build/kiali.mk):setup-kiali-multicluster,run-evals-multicluster,redeploy-kiali-multicluster-dev,kind-delete-multicluster. Uses Kiali dev/master (KIALI_MC_VERSION=dev) forlist_clusters. CI fix:SHELL := /bin/bashforpipefailrecipes.run-evaluation-multiclusterjob when suite iskiali: spins up east/west Kind, starts MCP with multicluster config, runs the 6 tasks.docs/KIALI.mdupdated with setup/run instructions and troubleshooting.** Requires the changes in #1224 to work