janhoy opened a new pull request, #904: URL: https://github.com/apache/solr-operator/pull/904
Two fixes to the e2e harness, found while running the full suite on two Kind clusters at once (related to #901 / #903). - **Pin everything to the test cluster's context.** `kubectl`, `helm` and the Go suite used the shared current-context, which `kind create cluster` switches. Two parallel runs (e.g. two Solr versions) both ended up on whichever cluster was created last. They now use `KUBE_CONTEXT` (`--context`, `HELM_KUBECONTEXT`, `GetConfigWithContext`). - **Pull the largest images once and share them with the Kind nodes.** Each node pulled every image itself, so a run downloaded the Solr image three times, and on a slow network the first specs timed out waiting for it. The Solr, ZooKeeper and rawfile-localpv images (the only ones over 100MB) are now pulled once on the host, in parallel with cluster creation, and imported into each node. Images that are already local are not pulled again. Failures fall back to the nodes pulling as before. Tested the image caching on macOS (bash 3.2) and Linux (bash 5), including that pods run the imported images with `imagePullPolicy: Never` (also the amd64-only ZooKeeper image on arm64). 🤖 Done 100% by Claude Code Opus -- 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]
