[
https://issues.apache.org/jira/browse/IGNITE-14977?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Sergey Chugunov updated IGNITE-14977:
-------------------------------------
Description:
Raft supports some kind of rebalancing out-of-the-box by sending pieces of its
log from up-to-date nodes to a fell behind nodes.
But when there is not enough Raft log available (because Raft took a snapshot
in storage and compacted its log) Raft tries to send a snapshot first.
Storage should provide an API to Raft to send a snapshot to a given node.
was:
Raft log of each partition (Raft group) in Ignite 3 acts like a WAL log from
Ignite 2.x.
Thus persistent store doesn't need to maintain its own WAL log but should
integrate with Raft log for purposes of crash recovery.
It is expected that integration will use Raft revision as an analogue of
partition counter from Ignite 2.x so API for access to revision is needed.
> Persistent store and Raft integration for rebalancing
> -----------------------------------------------------
>
> Key: IGNITE-14977
> URL: https://issues.apache.org/jira/browse/IGNITE-14977
> Project: Ignite
> Issue Type: Task
> Reporter: Sergey Chugunov
> Priority: Major
> Labels: iep-74, ignite-3
> Fix For: 3.0.0-alpha3
>
> Original Estimate: 144h
> Remaining Estimate: 144h
>
> Raft supports some kind of rebalancing out-of-the-box by sending pieces of
> its log from up-to-date nodes to a fell behind nodes.
> But when there is not enough Raft log available (because Raft took a snapshot
> in storage and compacted its log) Raft tries to send a snapshot first.
> Storage should provide an API to Raft to send a snapshot to a given node.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
