[jira] [Commented] (DRILL-1343) Drill should timeout after short time if a storage plugin is unresponsive.

["Sean Hsuan-Yi Chu (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/DRILL-1343?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14225841#comment-14225841 ] 

Sean Hsuan-Yi Chu commented on DRILL-1343:
------------------------------------------

This issue happened on Impala as well. See https://issues.cloudera.org/browse/IMPALA-1031. Essentially, it takes HBase/Zookeeper a long time to fail if we choose default retry times and timeout.

A solution could be:
In bootstap-storage-plugins for hbase, add these:
        "zookeeper.session.timeout": "60000",
        "zookeeper.recovery.retry": "0",
        "hbase.client.retries.number": 1

According to my test, it works. But I am wondering if this could cause subtle side-effect towards Drill?

> Drill should timeout after short time if a storage plugin is unresponsive.
> --------------------------------------------------------------------------
>
>                 Key: DRILL-1343
>                 URL: https://issues.apache.org/jira/browse/DRILL-1343
>             Project: Apache Drill
>          Issue Type: Bug
>            Reporter: Zhiyong Liu
>            Assignee: Sean Hsuan-Yi Chu
>            Priority: Blocker
>             Fix For: 0.7.0
>
>         Attachments: DRILL-1343.1.patch
>
>
> git.commit.id.abbrev=654c879
> git.commit.id=654c879f7caa13925edca911de1b59d04d8f1a8b
> Start drillbit and sqlline with a schema specified, e.g.,
> sqlline -n admin -p admin -u "jdbc:drill:schema=dfs.TpcHMulti;zk=10.10.30.104:5181,10.10.30.105:5181,10.10.30.106:5181"
> Execute one of the following:
> show tables;
> select * from INFORMATION_SCHEMA.`TABLES`;
> The commands hang forever.  No exception was thrown in the log file.
> Note that if using zk=local, the second query works with no hanging problems.



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