I’m planning to use FailFastStrategy and getAnyReplica afterwards to handle data retrieval failure but I have a concern about this scenario;
- User update data
- SDK upsert data using “Majority” durability options to a bucket with 2 replica
- Replication to replica 1 is succeeded
- SDK receives success response
- Replication to replica 2 is failed/delayed
- User retrieve data
- Request to active node is failed and immediately return failed response due to FailFastStrategy
- SDK try to get from replica using “getAnyReplica”
- It tries to get data from replica 2 which not yet updated
- SDK receives outdated data
Is above scenario is possible…?
If yes, is there any way to guarantee we can get data from updated replica…?