CPS-2000 Schema object cache not distributed.

References

CPS-2000 - Getting issue details... STATUS

Issues & Decisions

Solution Proposal

Assumption

1. during Upgrade two CM Handle LCM notifications are sent: 1) READY→LOCKED (for upgrade). 2) LOCKED → READY
2. LCM events should be processed within Milliseconds

Alternative 0: Remove existing cache

• tried but mush slower and ODL memory leak experienced

Alternative 1: (existing) Kafka Message without confirmation

Notes

1. LCM event should; be processed within milliseconds
2. Caffeine cache has an expiry (10mins) So NOT all models are cached all the time
3. Worst case scenario: an update (with newly modelled data) happens before cache expiry. Update gets reject (fails validation)
   1. mitigation: Invalidate cache upon validation error and try again (consider this as an extension alternative 1a )

Alternative 2: (existing) Kafka Message with (new) Kafka Confirmation

Not viable without knowing how many instances need to form ie need Hazelcast Cache

Alternative 3: (existing) Kafka Message with (new) Hazelcast Cache (semaphore)

Confirmation

Alternative 4: (existing) Polling (status watchdog) with (new) confirmation

Not viable because upgrades can be completed well withing polling time

Alternative 5: Add version info the cache-key