Geo Test Audit: review epics
This issue summarizes a review of recent epics related to Geo, with the goal of identifying any common themes related to testing needs.
The focus was on epics here because they already group issues based on common themes.
I will close this issue when the main conclusions from this review are captured in future issues or document.
These groupings don't cover all epics or themes
Epics:
- Admin experience improvements (&91 (closed)). Can testing help improve/increase confidence in the current experience?
- Improve uploads support &1180 (closed)
- Simplify install process
- Assistance debugging setup
- Resolve setup problems (issues mainly documentation...a lot of the process is not automated(?))
- Simplify upgrade process (&1451 (closed) still being fleshed out, but could involve increased automation of steps, which will warrant testing)
- Data handling/management
- Replicated data verification
- Expand replicated data types
- Data housekeeping
- Improved object storage support
-
Promotion, demotion, failover (already created a separate issue to investigate this further)
-
(Non-DR) node management (new capabilities not yet functional, no testing needs right now)
- Secondary mimicry/primary nodes behind load balancer
- Automatically choose best node for user experience
- Other epics (do these fit under previous groups?)
- Selective Sync Enhancements (awaiting further demand)
- Hashed storage (rollout in progress)
- GDK support
- Omnibus Improvements, FDW
Question: a lot of admin tasks appear to be manually done (and not necessarily via the UI). How often are these processes currently QA'd? Are there regularly planned run-throughs of these manual tasks and/or documented manual QA procedures? Should we define these various manual workflows and assess how they are covered by existing tests?
Our E2E tests are UI-driven, but it seems where test/quality support is most needed/more directly associated with the team's current work are lower level testing (e.g. of new/improved data management processes), test infrastructure/setup, perhaps manual testing/documentation of multi-step admin tasks.
New tests will be needed as these tasks become more automated or UI-driven.
I'll create a separate issue to assess if the current E2E tests cover From an end-user (non-admin user) standpoint.
I'll also create an issue to review this summary of issues related to testing from the backlog.