Migrate Backend to HA Redis minds/infrastructure/infrastructure#46
Ticket(s) / Related Merge Requests
Issue: minds/infrastructure/infrastructure#46 (closed)
Summary of Changes
This MR introduces support for using the engine with a highly available Redis cluster. This allows for us to use a single hostname for HA proxy and serve the Redis master and slave on different ports, while allowing for the old structure to be used as well.
Testing Considerations
Sandbox is currently using the new Redis deployment, you can view the Redis pods in the caching
namespace:
kubectl get po -n caching
You can also run the Helm tests in order to ensure that the NetworkPolicy
resources are working correctly for the engine. Note that the runners tests will fail in sandbox.
helm test infra-redis-ha-46
Deployment Considerations
This change requires modifications to the structure of the redis
section within our settings.php
. As we share a ConfigMap
across environments, we will likely need to deploy a temporary revision and direct staging and canary at this revision. Finally, we can make another MR on the Helm chart to migrate fully.
After we're happy that the new Redis is functioning as expected, we should come back and migrate fully to the new structure.
Regression Scope
Caching
Platform Affected (web, mobile, etc)
Both
Developer Testing Completed
Tested using the engine in sandbox with the new Redis
Screenshots / Screen Recording
N/A
Does this impact
-
Localization -
Dark/light mode -
Guest mode
Definition of Done Checklist
-
The Acceptance Criteria has been met -
Code is tested: Testing includes unit/spec, E2E/automated and manual testing -
Merge requests description has been filled out