You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@cassandra.apache.org by "Sylvain Lebresne (JIRA)" <ji...@apache.org> on 2016/06/07 15:58:21 UTC

[jira] [Commented] (CASSANDRA-11963) Paged queries limited to Integer.MAX_VALUE total rows

    [ https://issues.apache.org/jira/browse/CASSANDRA-11963?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15318744#comment-15318744 ] 

Sylvain Lebresne commented on CASSANDRA-11963:
----------------------------------------------

You are right, and I guess it's some form of oversight. {{Integer.MAX_INT}} sounded "good enough for everyone" on first approximation (read, for any query that is meant to be somewhat real-time), but I certainly can see how you could blew it out if you iterate over a large chunk of your database.

Unfortunately, I'm not sure how feasible it is to fix this in 2.1/2.2: we can't change the format of the paging state without breaking backward compatibility and are thus stuck to 4 bytes. Fortunately, for 3.0, we have changed the paging state format and use vints which will allow use to transparently pass a long. Given this, and given that this is the first time someone complains about this (as in, I don't think people are hitting this limitation all the time), I suggest we fix this in 3.0+ and just document it as a limitation in prior versions. 

> Paged queries limited to Integer.MAX_VALUE total rows
> -----------------------------------------------------
>
>                 Key: CASSANDRA-11963
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-11963
>             Project: Cassandra
>          Issue Type: Bug
>          Components: Core
>            Reporter: Adam Holmberg
>            Priority: Minor
>             Fix For: 2.1.x, 2.2.x
>
>
> Paged queries are artificially limited to Integer.MAX_INT rows in total. This appears to be related to PagingState.remaining, which decrements monotonically as pages are consumed. 
> I don't think this is intentional behavior, and haven't found any mention of it in the docs.
> Issue observed in latest 2.1 and 2.2 releases. Does not occur in 3.x



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