Investigate additional mitigations to cache invalidation issues
If Redis goes down, Pulp can no longer invalidate / clear out the cache appropriately while it is offline, and when it comes back up it will be out of date.
Additionally there could be smaller transient errors that prevent invalidation events from going through.
The time window this damage can manifest is pretty small (10 minutes as of #8996) but we should investigate further methods of mitigation. One example could be - pulpcore detecting when the Redis connection has failed, and either retrying cache invalidations, or flushing the cache entirely.
Please register to edit this issue