[jira] [Commented] (CASSANDRA-14793) Improve system table handling when losing a disk when using JBOD

["Benjamin Lerer (Jira)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/CASSANDRA-14793?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17119639#comment-17119639 ] 

Benjamin Lerer commented on CASSANDRA-14793:
--------------------------------------------

{quote}
Of course, if the tables in system could be regenerated, that would be better still but I'm not sure what that looks like complexity-wise versus pinning it.
{quote}

Regenerating the system tables is far more complex that simply pinning the system tables to a given disk and by consequence more risky.

I made an initial patch [here|https://github.com/apache/cassandra/compare/trunk...blerer:CASSANDRA-14793]. The patch allow to configure a specific directory for storing system keyspaces data (local and replicated) through the {{system_keyspaces_data_file_directory}} property in the cassandra.yaml file. This directory must be different from the other data directories to ensure that it cannot be filled by the data of the other keyspaces. If the {{system_keyspaces_data_file_directory}} property was not set the behavior was the same as the one we currently have.

Another approach suggested by [~marcuse] would be to pin the system keyspaces data to the first of the of the {{data_file_directories}}.

For both solutions we would need to find a way for the user to easily migrates it systems tables data into the new directories if needed.

I do not have a strong opinion on which solution is better and would like to get some feedback from people having more experience that me with large production clusters.

> Improve system table handling when losing a disk when using JBOD
> ----------------------------------------------------------------
>
>                 Key: CASSANDRA-14793
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-14793
>             Project: Cassandra
>          Issue Type: Bug
>          Components: Legacy/Core
>            Reporter: Marcus Eriksson
>            Assignee: Benjamin Lerer
>            Priority: Normal
>             Fix For: 4.0
>
>
> We should improve the way we handle disk failures when losing a disk in a JBOD setup
>  One way could be to pin the system tables to a special data directory.



--
This message was sent by Atlassian Jira
(v8.3.4#803005)

---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscribe@cassandra.apache.org
For additional commands, e-mail: commits-help@cassandra.apache.org