cassandra-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Murukesh Mohanan (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (CASSANDRA-12244) progress in compactionstats is reported wrongly for view builds
Date Fri, 11 Nov 2016 04:45:58 GMT

    [ https://issues.apache.org/jira/browse/CASSANDRA-12244?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15656122#comment-15656122
] 

Murukesh Mohanan commented on CASSANDRA-12244:
----------------------------------------------

[~jasonstack] IMHO the ranges don't need to be sorted. You could do something like:

{code}
        for (Range<Token> range : StorageService.instance.getLocalRanges(baseCfs.keyspace.getName()))
        {
            rangesTotal++;
            // Each range whose right boundary (largest value) is less than the lastToken
can be considered as seen
            // by the node.

            if ((lastToken != null) && lastToken.compareTo(range.right) > 0)
                rangesCompleted++;
        }
{code}

Making a collection and sorting it is a waste when you need to iterate over the ranges only
once.

> progress in compactionstats is reported wrongly for view builds
> ---------------------------------------------------------------
>
>                 Key: CASSANDRA-12244
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-12244
>             Project: Cassandra
>          Issue Type: Bug
>            Reporter: Tom van der Woerdt
>            Assignee: ZhaoYang
>            Priority: Minor
>              Labels: lhf
>             Fix For: 3.0.9
>
>
> In the view build progress given by compactionstats, there are several issues :
> {code}
>                                      id   compaction type   keyspace         table  
completed       total     unit   progress
>    038d3690-4dbe-11e6-b207-21ec388d48e6        View build      mykeyspace   mytable 
 844 bytes   967 bytes   ranges     87.28%
> Active compaction remaining time :        n/a
> {code}
> 1) those are ranges, not bytes
> 2) it's not at 87.28%, it's at ~4%. the method for calculating progress in Cassandra
is wrong: it neglects to sort the tokens it's iterating through (ViewBuilder.java) and thus
ends up with a random number.



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

Mime
View raw message