cassandra-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Brandon Williams (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (CASSANDRA-6257) Safety check on node joining cluster (based on last seen vs GC grace period) to avoid zombie data
Date Tue, 29 Oct 2013 20:45:25 GMT

    [ https://issues.apache.org/jira/browse/CASSANDRA-6257?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13808429#comment-13808429
] 

Brandon Williams commented on CASSANDRA-6257:
---------------------------------------------

Not exactly, because it's hard to discern what part of the cluster means locally.  I guess
you could say 'last time I saw another node alive' but even then in the case of a long network
split you might exceed gc grace.

> Safety check on node joining cluster (based on last seen vs GC grace period) to avoid
zombie data
> -------------------------------------------------------------------------------------------------
>
>                 Key: CASSANDRA-6257
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-6257
>             Project: Cassandra
>          Issue Type: Improvement
>            Reporter: Johnny Miller
>            Priority: Minor
>
> When a node is rejoining a cluster, it would be nice to have some form of safety check
that the cluster recognises the last time the node was part of the cluster is greater that
the GC grace period and therefore should not be able to rejoin the cluster unless the administrator
specifically requests it.
> The goal of this is to help avoid the potential issues with deleted data coming back
from the rejoining nodes dataset.



--
This message was sent by Atlassian JIRA
(v6.1#6144)

Mime
View raw message