[
https://issues.apache.org/jira/browse/IGNITE-17638?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Alexander Lapin resolved IGNITE-17638.
--------------------------------------
Resolution: Duplicate
> Implement Txn state map to optimize write intent resolution
> -----------------------------------------------------------
>
> Key: IGNITE-17638
> URL: https://issues.apache.org/jira/browse/IGNITE-17638
> Project: Ignite
> Issue Type: Improvement
> Reporter: Sergey Uttsel
> Priority: Major
> Labels: ignite-3, transaction3_ro
>
> Need to implement Txn state map which contains txId -> (txState,
> txCoordinatorAdress, commitTimestamp).
> # This map will be located on all nodes in replication group (primary and
> followers) in PartitionReplicaListener and PartitionListener.
> # Updated on ReadWriteReplicaRequest receiving in PartitionReplicaListener.
> Initially, only txCoordinatorAddress is known. The coordinator address is
> available in ReplicaManager on receiving a ReplicaRequest. Need to pass it to
> PartitionReplicaListener.
> # PartitionReplicaListener use raftClient commands for replication. So need
> to send coordinator address with raft command and save it in
> PartitionListener.
> # On TxCleanupReplicaRequest receiving need to set txState and
> commitTimestamp (if commited).
>
> Txn state map optimizes write intent resolution. If commitTs <= readTs then
> RO transaction treat a write intent as committed.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
