Our documentation should contain a dedicated page on our possible upgrade scenarios.
Upgrades of the platform can be done in multiple steps:
- Optional: Set
reconciliationPaused to true for some or all of your products
- Update CRDs
- Update Operators
- Optional: Set
reconciliationPaused to false
Then we also have the product updates
There are a few extra things we should document or take into consideration:
- Extra steps/caution needed when using custom images
- Document that (and why) pods will restart after an operator upgrade
- Document the fact that Helm can't manage CRDs: https://helm.sh/docs/chart_best_practices/custom_resource_definitions/
- Document OpenShift upgrade procedure
- Hint at platform release notes
- Hint at always having to check the release notes of the underlying products