• Home
  • Features
  • Pricing
  • Docs
  • Announcements
  • Sign In

chideat / valkey-operator / 36516369749
35%
main: 36%

Build:
Build:
LAST BUILD BRANCH: docs-artifacthub-badge
DEFAULT BRANCH: main
Ran 29 Sep 2026 03:19AM UTC
Jobs 1
Files 151
Run time 1min
Badge
Embed ▾
README BADGES
x

If you need to use a raster PNG badge, change the '.svg' to '.png' in the link

Markdown

Textile

RDoc

HTML

Rst

29 Sep 2026 03:15AM UTC coverage: 35.441% (+0.009%) from 35.432%
36516369749

push

github

web-flow
ci(e2e): run the whole suite in the cluster, one job per Valkey version (#153)

The workflow could not run the suite it dispatches:

- The operator image was tagged "e2e". The chart hands the operator its
  image tag as its version, and the operator picks the actors that run
  each instance by it; "e2e" is not a semantic version, so no instance got
  past its first command and no pod was ever created.
- A single-node kind cluster cannot schedule e2e instances: they keep a
  shard's pods on different nodes (required anti-affinity), up to three
  pods per shard.
- The suite ran on the runner and dials the addresses in each instance's
  status; for ClusterIP instances those are pod IPs the runner cannot
  reach.
- The whole suite does not fit in the 6 hours a hosted runner allows. The
  old defaults skipped the failover and replication tests and Valkey 8.0
  and 8.1.

A planning job now lists the suite's own supportedVersions minus
skip_versions, and each version gets a job: a 3-node kind cluster
(config/kind/kind-e2e-config.yaml, control plane untainted, etcd in tmpfs),
the operator image tagged with the chart's appVersion ("latest" on main),
every image the suite deploys preloaded, and the suite running as a Job
built from Dockerfile.e2e. The version-independent specs (webhook,
operator) run in the first job; the TLS specs stay with the newest
version.

Ginkgo's 5-hour timeout fires before Go's and before the Job's deadline,
so a slow run still ends with its report; a container that has not
started within 5 minutes fails the job at once. The Job's own condition
decides the result, and each job prints ginkgo's summary from the pod's
log and uploads the logs.

The inputs keep their names but default to the whole suite; label_filter
narrows each job, and node_image picks the Kubernetes version. kind moves
to v0.33.0, its checksum verified, and cert-manager to v1.21.2. Workflow
inputs reach the scripts through the environment.

On release-2.1, where the op... (continued)

7768 of 21918 relevant lines covered (35.44%)

0.4 hits per line

Jobs
ID Job ID Ran Files Coverage
1 36516369749.1 29 Sep 2026 03:19AM UTC 151
35.44
GitHub Action Run
Source Files on build 36516369749
  • Tree
  • List 151
  • Changed 1
  • Source Changed 0
  • Coverage Changed 1
Coverage ∆ File Lines Relevant Covered Missed Hits/Line
  • Back to Repo
  • 5ee5470d on github
  • Prev Build on release-2.1 (#36411931026)
  • Next Build on release-2.1 (#44106115115)
STATUS · Troubleshooting · Open an Issue · Sales · Support · CAREERS · ENTERPRISE · START FREE TRIAL · SCHEDULE DEMO
ANNOUNCEMENTS · TWITTER · TOS & SLA · Supported CI Services · What's a CI service? · Automated Testing

© 2026 Coveralls, Inc