-
Type: Improvement
-
Resolution: Unresolved
-
Priority: Major - P3
-
None
-
Affects Version/s: None
-
Component/s: None
-
Cluster Scalability
-
0
The ReshardingTxnCloner is very conversvative when transferring retryable writes/transaction information from the donor. It just scans through the config.transactions table and convert them to IncompleteTransactionHistory records without attmepting to scan through the oplog history chain. This has a side effect of generating too many IncompleteTransactionHistory records on the recipients even though these transactions/retryable writes were originally from a different namespace.
This optimization has to also take into account special collections like time series (which has a different namespace name for the bucket collection)
- is related to
-
SERVER-92420 Resharding oplog application shouldn't use a unique lsid for each delete operation
- Closed