* Run e2e tests in k3s + helm rather than docker
* collect logs from k3s, use head and newer kube version, check for webhook up when starting rancher
* use latest rc chart
* debugging
* use lower kube version
* fixes to brittle tests, log fixes, enable oidc provider
* fix logs
* test tweaks
* use alpha repo, rc in stable is too old. also wait for capi service
* improvements
* debug
* rtryyr
* don't run settings e2e tests - they leave cluster with non-original values
* fix settings tests, fix CRD test (worked by luck before)
* fix fetch of initial settings
* more settings fixes, crd tweak
* update settings
- it uses useDefault as the reset, which was the issue
- leave tests as is (mosty) and just reset them all at the end
* asd
* adsad
* more settings fixes
* remove problem code
* use toms debug image for api failures
* sdfdsf
* Use 'head' chart
* Use audit log
* enable browser logs for beforeAll and afterAll
* debug for checking racnher pod containers, increase audit log level
* don't overwrite logs...
* output improvements, move position of audit log `set`
* use new (docs pending) flag to enable audit log
* extra logging now in rancher/rancher, use head image (should remove a lot of false positives around fleet)
* improve comment
* combine test artifacts
* tweaks, and another data point for 'resolved' tests
* attempt to get previous logs (audit log previous might not exist)
* remove k logs --previous, use logs deployment/name
* re-enable all tests
* tidyng up
* better logs, including crashed containers
* k3s updates... but run in docker
* change lookup from anything containing 1 to `1`
* changes following pr review
* remove un-needed chrome pin
- this commented out section was a c+p from the main e2e section
- the main e2e section comments this out
- both should use a single repo gh action, won't do that now though
* WIP - just testing how are suite behaves with vai on
* Use common method to create query params for steve api requests
* with vai on there's no workload page... so update tests
* flip back hom page urls
* more fixes
* More fixes
* E2eRequestUtils --> RequestUtils, smarter isVaiCacheEnabled
* better method name, fix a11y tests
* disable vai for vai tests
* fetching the setting gives a 404. hardcode for the moment
* give up, make method async
* use pathWithDefaultSteveParams everywhere
* rename grep tag vai to noVai (matching what the tests run)
* fixes
* Allow more time for rancher to restart after toggling a ff
* fix network policies
* Changes following review
* replace pathWithDefaultSteveParams with ?*
* fixes
* Remove temp code that enables FF, should be defunct once enabled by default
* unpin, improve script
* on failure upload rancher logs to job artifacts
* fix
* test specific rancher commit, and rancher debug level
* try head again following overnight fixes
* Wire in VAI_ENABLED env var (to avoid), fixed isVaiCacheEnabled (no value, use default)
* Standardise references to small image, use `k8s.gcr.io/pause` instead of `nginx:alpine`
* test - see if specific workalods are the sticking point
* reduce impact of cronjob
* reduce impact of cronjob - additional
* Improve e2e tests that spam resource creation/delete
- take gaps when creating resources
- wait for ns deletion (containing resources) to complete before continueing)
* even stricker cron schedule
* update three additional places where workloads spam happens
* fix tag of new @vai tests
* use default count, reduce to min. also removed horrible `while`
* temp fail cronjobs to see logs + recordings
* improvements to noVai test cases, move new utils to commands
* fix flaky compliance and kontainer tests
* proper context, fix lint, fix ingress test
* fix lint
* update rancher image version for e2e tests
* update package.json's docker local script to use the latest version
* update what's new version to 2.10
* revert unrelated change
* add rancher version variable to reuse + skipping some e2e tests
* revert unnecessary changes in version.spec.ts
* remove typo
* fix version typo
* skip account-api-keys tests
* skip a test in namespace-picker.spec.ts
* update home test to use rancher version variable
* remove 'v' from version number
* fix typo
* fix flaky charts tests
* fix flaky repositories tests
* more precise enabling logic for checkboxes
* lint
---------
Co-authored-by: Mo Mesgin <mmesgin@Mos-M2-MacBook-Pro.local>
- given we now serve the (built) dashboard files we set CATTLE_UI_OFFLINE_PREFERRED=true
- this means the embedded rancher/ui is used
- so now we do something similar to ensure the latest rancher/ui is used
* e2e: Add built UI to docker container rather than run a node process
* Fix to ensure docker:local:stop tried to both clean and remove cypress container
* Check dist folder is correct
* Build U first
* Remove debug, wait for ui
* Change cluster test
* Tweak test
* Improve resilience of cluster manager delete tests
* Fix lint
* Try and fix delete tests
* Remove unused var
* Remove unused baseUrl
* Fix cluster delete tests
* Address lint issues
* Fix api keys test
Richard Cox
Alejandro Ruiz
Alexandre Alves
Richard Cox
yonasberhe23
Phillip Rak
momesgin
Neil MacDougall
Richard Cox