bottlerocket-os / bottlerocket-os/bottlerocket-test-system
cluster preconditions check
- Dominant language
- Rust
- Stars
- 18
- Forks
- 27
- PR merge metrics
- No merged PRs in 30d
Description
As described in #428, having errant nodes in the cluster would produce incorrect conformance testing results. Additionally, if a conformance testing process failed without proper cleanup, it is possible for a testing cluster to have namespaces or other objects left behind that could cause conformance testing to fail.
We should probably do certain checks of a cluster's state before running conformance testing to ensure that we
- Have only the nodes we expect
- Are not going to fail due to the existence of certain namespaces in the cluster
This is probably applicable to ECS clusters for the same reasons.
Probably we want to do this in some combination of existing or new resource and/or test agents.
Contributor guide
Research direction
Start by reading issue #428 and locating the existing resource and test agents in the repository. Determine how cluster state is currently inspected before conformance tests, including node membership and leftover namespaces or objects, and compare the needs of Kubernetes and ECS clusters. Done means the preconditions are defined, implemented, and covered by tests for invalid cluster state.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- kubernetes, rust
- Domain
- infrastructure, testing-qa
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100