[
https://issues.apache.org/jira/browse/IGNITE-17477?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Vyacheslav Koptilin updated IGNITE-17477:
-----------------------------------------
Description:
After we have implemented a replication layer, a part of the RAFT command are
become useless: _GetAndDeleteCommand, UpsertCommand, GetAllCommand, GetCommand,
DeleteExactCommand_ (the list can be changed) and another one required
modification, because all raft command should apply _rowId_ and never try to
match some row to its id (it is already done by replication layer).
Also required to extract a primary index (for now, it is a map) from the RAFT
state machine. It will be used by relocation layer for read, but in the state
machine will use it for modification only.
was:
After we have implemented a replication layer, a part of the RAFT command are
become useless: _GetAndDeleteCommand, UpsertCommand, GetAllCommand, GetCommand,
DeleteExactCommand_ (the list can be changed) and another one required
modification, because all raft command should apply _rowId_ and __ never try to
match some row to its id (it is already done by replication layer).
Also required to extract a primary index (for now, it is a map) from the RAFT
state machine. It will be used by relocation layer for read, but in the state
machine will use it for modification only.
> Redesign RAFT commands in according with replication layer
> ----------------------------------------------------------
>
> Key: IGNITE-17477
> URL: https://issues.apache.org/jira/browse/IGNITE-17477
> Project: Ignite
> Issue Type: Improvement
> Reporter: Vladislav Pyatkov
> Priority: Major
> Labels: ignite-3
>
> After we have implemented a replication layer, a part of the RAFT command are
> become useless: _GetAndDeleteCommand, UpsertCommand, GetAllCommand,
> GetCommand, DeleteExactCommand_ (the list can be changed) and another one
> required modification, because all raft command should apply _rowId_ and
> never try to match some row to its id (it is already done by replication
> layer).
> Also required to extract a primary index (for now, it is a map) from the RAFT
> state machine. It will be used by relocation layer for read, but in the state
> machine will use it for modification only.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
