You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@ignite.apache.org by "Aleksandr Polovtcev (Jira)" <ji...@apache.org> on 2022/11/16 09:36:00 UTC

[jira] [Updated] (IGNITE-18172) Raft learners rebalance

     [ https://issues.apache.org/jira/browse/IGNITE-18172?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Aleksandr Polovtcev updated IGNITE-18172:
-----------------------------------------
    Description: 
Currently, the Rebalance protocol only applies to Raft followers. We would like to extend it to be applicable to rebalancing Raft learners as well. The following should be done:

# {{partition.assignments.*}} Metastorage keys should store followers' assignments as well as learners' assignments. I propose adding a flag to the corresponding entries (i.e. {{isFollower}}).
# When reacting to changes in Metastorage entries, call a new version of `changePeersAsync` (see IGNITE-18155) that will provide new followers and learners.

Concrete applications of learners rebalance and actual learner assignments are subject of further design (e.g. it can be used in Metastorage and Replicated Tables).  

> Raft learners rebalance
> -----------------------
>
>                 Key: IGNITE-18172
>                 URL: https://issues.apache.org/jira/browse/IGNITE-18172
>             Project: Ignite
>          Issue Type: Task
>            Reporter: Aleksandr Polovtcev
>            Priority: Major
>              Labels: ignite-3
>
> Currently, the Rebalance protocol only applies to Raft followers. We would like to extend it to be applicable to rebalancing Raft learners as well. The following should be done:
> # {{partition.assignments.*}} Metastorage keys should store followers' assignments as well as learners' assignments. I propose adding a flag to the corresponding entries (i.e. {{isFollower}}).
> # When reacting to changes in Metastorage entries, call a new version of `changePeersAsync` (see IGNITE-18155) that will provide new followers and learners.
> Concrete applications of learners rebalance and actual learner assignments are subject of further design (e.g. it can be used in Metastorage and Replicated Tables).  



--
This message was sent by Atlassian Jira
(v8.20.10#820010)