When running this test against an AWS cluster, this operation was resulting in 504s. This was potentially hitting some AWS rate limits when making these requests concurrently.
* fix(e2e): Address flakiness during screen navigation
* fix(e2e): Fix function call
* fix(e2e): Address flakiness in connection
* fix(e2e): Make locator explicit to avoid conflict when an ID has `ok`
* fix(e2e): Fix URL now that project page defaults to showing targets
* test(e2e): decouple url used for browser vs cli
* test(e2e): Add instructions for using the UI of a local branch
* CR
* Fix typo in e2e-tests/README.md
Co-authored-by: Dharini Jeeva <dharini@hashicorp.com>
---------
Co-authored-by: Dharini Jeeva <dharini@hashicorp.com>
This test attempted to delete a session recording when the policy did not allow it. However, there was a UI change to remove that option entirely, so we are removing the check.
* test(e2e): Update "Inactive" locators
* test(e2e): Update worker filter locators
* test(e2e): Address flake
Ensure that the test has reached the org page before trying to click the Auth Methods link (there is an auth method link on both the global page and the org page).
* test(e2e): Address deprecation
* test(e2e): Add additional connection assertions on tests
* CR: Fix flake when waiting for sessions
If no sessions are available and the page needs to reload, we should wait to ensure that the page has loaded before checking what the state of the sessions are
* CR: Address flake in connect methods
Noticed an issue where a session recording was not playable. The theory is that the session didn't last long enough before it was cancelled. This adds a 2 second wait after establishing a connection before continuing.
* CR: Use authenticated state by default
* CR: Update dynamic host catalog assertion
* CR: Simplify host set checks
- Use .all() when looping
- Add check to see if we need to reload the page or not
* CR: Be more specific with locator
Ran into a conflict with the Static/Dynamic host catalog type and the Static/Dynamic credential type.
* CR: Update connection methods to be more resilient
Rather than using a hardcoded sleep, we will now run a command on the target and wait for a specific string to return before continuing.
* CR
* test(e2e): Update locators that can't be found
* test(e2e): Fix race condition
It seems like the alias tests were failing as it was unable to open the Manage menu here. One theory is that we need to make sure the alias page has fully loaded before it tries to click the Manage button (the test might be finding the Manage button from the previous screen and clicking that)
* CR: Fix Alias header check
* refact(e2e): Split up boundary-cli helper methods
* CR: Use ESM import
* CR: Move authenticated state to a non-shared file
* CR: Use mjs
* CR: Modify playwright config to .mjs
* test(e2e): Only clean up org if org hasn't been deleted yet
* test(e2e): Use new worker filter tab
* test(e2e): Update selector for updated Manage button
Some pages now use a Manage button instead of a Manage link. The "Title" selector was not being picked up on the Manage button.
* test(e2e): Fix flakiness due to race condition
The test should navigate to the home page after creating all of the resources using the API
* test(e2e): Add cleanup to test
* test(e2e): Fix flakiness when selecting links
Some links exist on both the "Global" and "Org" scope (i.e. Storage Policies, Auth Methods). There was some flakiness observed where the test intends to select the "Org" variant of the link, but the test moves too quickly and selects the "Global" variant instead.
This commit ensures that the Org page has loaded after clicking the Org link before trying to navigate to the next thing.
* test(e2e): Address flake due to OK conflict
* test(e2e): Address flake due to OK conflict part 2
* CR: Fix more race conditions
* test(e2e): Add check for multiple host sets
* CR: TCP targets don't have injected credentials
* test(e2e): Add "exact: true" for OK button locators
Sometimes, the OK button will get interfered by resource ids that contain an OK.
* test(e2e): Update password test to log in with new account
* CR: Don't break other tests
Performing a log out in a test when using authenticatedState causes subsequent tests that use the state to fail (it will show up as logged out)
* CR: Only do clean up steps when necessary
* CR: Use const
atlassian/gajira-* actions are no longer maintained, but they are using node16, which is soon to be deprecated. This change removes the use of those actions.
When running the Admin UI e2e tests against 0.16.0, there were some issues with trying to click on a link on the new pagination tables (links were not seen as visible). For some reason, updating the playwright version fixed this.
* test(admin/e2e): Modify password locators due to new Show Password button
* test(admin/e2e): Update breadcrumbs locators due to design updates
Previously, all breadcrumb items were links, but that is no longer the case for the active breadcrumb.
* test(admin/e2e): Update locator for worker filter textbox
* test(admin/e2e): Update locator for the New button
* test(admin/e2e): Update locator for New link
* fixup! test(admin/e2e): Update locator for worker filter textbox
Adds an end-to-end test for the Admin UI using Playwright. This test uses the Admin UI to create the necessary resources to connect to a target. The test then uses the CLI to start a session and checks in the Admin UI that a session is created
Michael Li
