[ 
https://issues.apache.org/jira/browse/CASSANDRA-21129?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18053797#comment-18053797
 ] 

Abhijeet Dubey commented on CASSANDRA-21129:
--------------------------------------------

The changes proposed here are for truly offline scenarios where you need to 
inspect metadata from SSTables when a node cannot start at all.

 

Looking at the changes proposed for CASSANDRA-19151 seems like it's using 
`ClusterMetadataService.deserializeClusterMetadata()` which essentially reads 
from disk and creates the CM object. Whereas the changes proposed here are 
reading from SSTables directly and create the `DumpData` which contains:
* metadata: Final reconstructed ClusterMetadata (latest snapshot + applied 
transformations)
* logState: Contains base snapshot + list of transformation entries for 
--epochs display
* snapshotEpochs: List of epoch numbers where snapshots exist for --snapshots 
display
 
I think both the PRs are complementary and not competing. A complete solution 
might combine both approaches: SSTable reading capability (this change) with 
their modification/repair commands (your change)?

 

> Tool to offline-dump the cluster metadata from an instance
> ----------------------------------------------------------
>
>                 Key: CASSANDRA-21129
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-21129
>             Project: Apache Cassandra
>          Issue Type: New Feature
>          Components: Transactional Cluster Metadata
>            Reporter: Abhijeet Dubey
>            Assignee: Abhijeet Dubey
>            Priority: Normal
>         Attachments: text-output.txt, to-string-dump-output.txt
>
>          Time Spent: 10m
>  Remaining Estimate: 0h
>
> When a Cassandra node fails to start due to Transactional Cluster Metadata 
> (TCM/CEP-21) corruption or issues, operators need a way to inspect the 
> cluster metadata state offline without starting the node. The existing tools 
> (nodetool, cqlsh) require a running node, leaving operators blind when 
> debugging startup failures.
>  
> When a node fails to start due to TCM corruption or inconsistencies, 
> operators have no way to inspect the metadata state without a running node. 
> This tool fills that gap by reading directly from SSTables.



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

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to