How do you safely test automation that has write access to a cluster?
retained text
Thinking about agents or automation that can restart, scale or roll back workloads. Where do you test something like that safely? kind, a staging cluster, something else? And has anyone found a decent way to reproduce messy failures (network problems, noisy neighbors, bad config pushes) so the tests are repeatable?
document fetched 2026-10-29 00:39 · canonical https://reddit.com/r/kubernetes/comments/1xhigl7/how_do_you_safely_test_automation_that_has_write