bitflicker64 commented on PR #3132: URL: https://github.com/apache/hugegraph/pull/3132#issuecomment-5444904419
A note on CI scope, so the follow-up is written down somewhere. `helm-chart-ci.yml` covers the static layer: lint, template rendering with assertions on the output, the validateValues rejection cases, kubeconform, and the legacy `--reuse-values` path. That is about everything reachable without a cluster. Two gaps for a follow-up: 1. Nothing installs the chart, so CI never exercises the running system. A `values-single.yaml` (1+1+1) install fits `ubuntu-latest`. The constraint is wall clock rather than memory: Server has a 450s startup floor and the Store `wait-for-pd` init can run to 900s, so that job needs a 15 to 25 minute budget and a `timeout-minutes`, currently unset at job and step level. 2. Nothing asserts on data across more than one graph. The `helm test` hook only checks that curl exits zero, and CI never creates a release so it never runs. A minimal version would create two graphs, write to one, and assert the other is unaffected, including after a truncate. No fault injection needed, so it fits the same runner. Full fault scenarios (partition, quorum loss, pod IP churn) need the 3 PD + 3 Store + 3 Server topology and can land inconclusive, so they fit a scheduled run rather than a per PR gate. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
