cassandra-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "sankalp kohli (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (CASSANDRA-10023) Emit a metric for number of local read and write calls
Date Wed, 20 Jan 2016 23:19:39 GMT

    [ https://issues.apache.org/jira/browse/CASSANDRA-10023?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15109710#comment-15109710
] 

sankalp kohli commented on CASSANDRA-10023:
-------------------------------------------

Your patch has a bug which will always mark things as replica non aware. In StorageProxy,
there will always be endpoints which are not the machine itself. Say endpoints in SP.syncWriteToBatchlog
are A,B and C and the code is running in A, it will mark replicaAware once and replicaNotAware
twice. 

> Emit a metric for number of local read and write calls
> ------------------------------------------------------
>
>                 Key: CASSANDRA-10023
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-10023
>             Project: Cassandra
>          Issue Type: Improvement
>            Reporter: sankalp kohli
>            Assignee: Tushar Agrawal
>            Priority: Minor
>              Labels: lhf
>             Fix For: 3.x
>
>         Attachments: CASSANDRA-10023.patch
>
>
> Many C* drivers have feature to be replica aware and chose the co-ordinator which is
a replica. We should add a metric which tells us whether all calls to the co-ordinator are
replica aware.
> We have seen issues where client thinks they are replica aware when they forget to add
routing key at various places in the code. 



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message