[
https://issues.apache.org/jira/browse/CASSANDRA-21627?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18116119#comment-18116119
]
n.v.harikrishna commented on CASSANDRA-21627:
---------------------------------------------
Thanks [~marcuse], [~smiklosovic], [~samt] for the inputs and the review!
> Allow nodetool cms reconfigure to ignore specific hosts
> -------------------------------------------------------
>
> Key: CASSANDRA-21627
> URL: https://issues.apache.org/jira/browse/CASSANDRA-21627
> Project: Apache Cassandra
> Issue Type: Improvement
> Components: Transactional Cluster Metadata
> Reporter: n.v.harikrishna
> Assignee: n.v.harikrishna
> Priority: Normal
> Fix For: 6.0-alpha3, 7.0
>
> Attachments: ci_summary_6.0.html, ci_summary_trunk.html
>
> Time Spent: 10m
> Remaining Estimate: 0h
>
> When shrinking a cluster, decommissioning a CMS member moves that role to
> another node, which may itself be due for decommission. The CMS then has to
> be reconfigured again, and this can repeat many times during a single shrink.
> Proposing to add _--ignore_ flag to the cms reconfigure command (like cms
> initialize has), so that the nodes going to be decommissioned can be excluded
> from CMS responsibilities and CMS set remains stable during the operation.
> One reconfiguration up front is then enough. The CMS stays on the surviving
> nodes for the whole shrink, because decommissioning a node that is not a CMS
> member does not trigger a reconfiguration.
> The ignore list applies only to that command and is not stored, so nodetool
> cms describe keeps reporting the real state.
> Proposed command format:
> {{nodetool cms reconfigure <rf> --ignore <host> [--ignore <host> ...]}}
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
