[
https://issues.apache.org/jira/browse/IGNITE-27851?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Mirza Aliev updated IGNITE-27851:
---------------------------------
Description:
Problem:
Users are not aware that Cluster Management Group (CMG) and Metastorage (MS)
voting members must be configured during the cluster initialization phase. This
leads to production clusters being initialized with only 1 voting member,
requiring full cluster re-initialization to fix.
Current State:
- Documentation about configuring CMG/MS voting members exists in the
ignite-cli-tool reference page (cluster init command section)
- This is not discoverable when users are learning about cluster setup or
system groups
- Administrator's Guide mentions system groups but doesn't clearly explain
initialization requirements
Impact:
* Users initialize clusters with default single-node CMG/MS
* Scale up to multi-node clusters expecting automatic rebalancing
* Discover they cannot add nodes to CMG/MS without cluster re-initialization
* No graceful way to reconfigure system groups
* System groups recovery is designed for disaster scenarios, not normal
configuration
Requested Improvements:
* Administrator's Guide additions:
** Add prominent section explaining CMG/MS voting member configuration MUST
happen at initialization
** Explain the difference between voting members and learners in RAFT
** Clarify that system groups cannot be rebalanced after initialization
** Link to cluster init command reference
* Quick Start / Getting Started:
** Include CMG/MS configuration examples in initial cluster setup
** Recommend production best practices (3+ voting members)
* System Groups Recovery documentation:
** Clarify this is for disaster scenarios only
was:
Problem:
Users are not aware that Cluster Management Group (CMG) and Metastorage (MS)
voting members must be configured during the cluster initialization phase. This
leads to production clusters being initialized with only 1 voting member,
requiring full cluster re-initialization to fix.
Current State:
- Documentation about configuring CMG/MS voting members exists in the
ignite-cli-tool reference page (cluster init command section)
- This is not discoverable when users are learning about cluster setup or
system groups
- Administrator's Guide mentions system groups but doesn't clearly explain
initialization requirements
Impact:
* Users initialize clusters with default single-node CMG/MS
* Scale up to multi-node clusters expecting automatic rebalancing
* Discover they cannot add nodes to CMG/MS without cluster re-initialization
* No graceful way to reconfigure system groups
* System groups recovery is designed for disaster scenarios, not normal
configuration
Requested Improvements:
# Administrator's Guide additions:
*
** Add prominent section explaining CMG/MS voting member configuration MUST
happen at initialization
** Explain the difference between voting members and learners in RAFT
** Clarify that system groups cannot be rebalanced after initialization
** Link to cluster init command reference
# Quick Start / Getting Started:
*
** Include CMG/MS configuration examples in initial cluster setup
** Recommend production best practices (3+ voting members)
# System Groups Recovery documentation:
*
** Clarify this is for disaster scenarios only
> Improve documentation for CMG/MS configuration during cluster initialization
> ----------------------------------------------------------------------------
>
> Key: IGNITE-27851
> URL: https://issues.apache.org/jira/browse/IGNITE-27851
> Project: Ignite
> Issue Type: Improvement
> Reporter: Mirza Aliev
> Assignee: Igor Gusev
> Priority: Major
> Labels: ignite-3
>
> Problem:
> Users are not aware that Cluster Management Group (CMG) and Metastorage (MS)
> voting members must be configured during the cluster initialization phase.
> This leads to production clusters being initialized with only 1 voting
> member, requiring full cluster re-initialization to fix.
> Current State:
> - Documentation about configuring CMG/MS voting members exists in the
> ignite-cli-tool reference page (cluster init command section)
> - This is not discoverable when users are learning about cluster setup or
> system groups
> - Administrator's Guide mentions system groups but doesn't clearly explain
> initialization requirements
> Impact:
> * Users initialize clusters with default single-node CMG/MS
> * Scale up to multi-node clusters expecting automatic rebalancing
> * Discover they cannot add nodes to CMG/MS without cluster re-initialization
> * No graceful way to reconfigure system groups
> * System groups recovery is designed for disaster scenarios, not normal
> configuration
> Requested Improvements:
> * Administrator's Guide additions:
> ** Add prominent section explaining CMG/MS voting member configuration MUST
> happen at initialization
> ** Explain the difference between voting members and learners in RAFT
> ** Clarify that system groups cannot be rebalanced after initialization
> ** Link to cluster init command reference
> * Quick Start / Getting Started:
> ** Include CMG/MS configuration examples in initial cluster setup
> ** Recommend production best practices (3+ voting members)
> * System Groups Recovery documentation:
> ** Clarify this is for disaster scenarios only
--
This message was sent by Atlassian Jira
(v8.20.10#820010)