cassandra-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Paulo Motta (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (CASSANDRA-10130) Node failure during 2i update after streaming can have incomplete 2i when restarted
Date Mon, 19 Jun 2017 16:05:00 GMT

    [ https://issues.apache.org/jira/browse/CASSANDRA-10130?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16054255#comment-16054255
] 

Paulo Motta commented on CASSANDRA-10130:
-----------------------------------------

bq. What we can do here is adding isFullRebuild to markIndexBuilt to only set the index as
queryable if the method is invoked by a full index rebuild

Good call! This approach looks better to me than handling state in non mark-methods and will
still keep the current behavior.

bq. Adding a cassandra.yaml option seems like a good idea, I think we could address it as
an improvement in a separate ticket. 

Agreed. The most typical failures during index rebuild are:
a) node crash
b) file system failure

This patch already provides a good solution to A), and B) should in most cases stop the node
via the default {{disk_failure_policy: stop}}, so while it would be nice to provide more tools
to allow handling 2i rebuild failure more gracefully (via auto-rebuild or nodetool) I think
non-crash/stop cases will be pretty uncommon so we can address those in a separate improvement
ticket. BTW, relating to point 2, I think we should probably run exceptions during 2i rebuild
failure ({{logAndMarkIndexesFailed}}) via the {{JVMStabilityInspector}}. WDYT?

> Node failure during 2i update after streaming can have incomplete 2i when restarted
> -----------------------------------------------------------------------------------
>
>                 Key: CASSANDRA-10130
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-10130
>             Project: Cassandra
>          Issue Type: Bug
>          Components: Coordination
>            Reporter: Yuki Morishita
>            Assignee: Andrés de la Peña
>            Priority: Minor
>
> Since MV/2i update happens after SSTables are received, node failure during MV/2i update
can leave received SSTables live when restarted while MV/2i are partially up to date.
> We can add some kind of tracking mechanism to automatically rebuild at the startup, or
at least warn user when the node restarts.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)

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


Mime
View raw message