You are viewing a plain text version of this content. The canonical link for it is here.

Posted to issues@hbase.apache.org by "Joseph (JIRA)" <ji...@apache.org> on 2016/07/22 17:53:20 UTC

[jira] [Commented] (HBASE-16221) Thrift server drops connection on long scans

    [ https://issues.apache.org/jira/browse/HBASE-16221?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15389919#comment-15389919 ] 

Joseph commented on HBASE-16221:
--------------------------------

I don't like the Unit Test as it looks like it could be flakey. I could just directly test a scanner and seeing if the lastAccessTime was updated, do you think that'd be a better test?

> Thrift server drops connection on long scans
> --------------------------------------------
>
>                 Key: HBASE-16221
>                 URL: https://issues.apache.org/jira/browse/HBASE-16221
>             Project: HBase
>          Issue Type: Bug
>          Components: Thrift
>    Affects Versions: 2.0.0, 1.2.0, 1.3.0
>            Reporter: Ashu Pachauri
>            Assignee: Joseph
>         Attachments: HBASE-16221.patch
>
>
> Thrift servers use connection cache and we drop connections after hbase.thrift.connection.max-idletime milliseconds from the last time a connection object was accessed. However, we never update this last accessed time on scan path. 
> By default, this will cause scanners to fail after 10 minutes, if the underlying connection object is not being used along other operation paths (like put).



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