[jira] [Commented] (KYLIN-1013) Make hbase client configurations like timeout configurable

["Shaofeng SHI (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/KYLIN-1013?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14736290#comment-14736290 ] 

Shaofeng SHI commented on KYLIN-1013:
-------------------------------------

The error log is like:
{code}
[Kylin][ODBC 1.0(w) Driver]The REST query request failed, the error message is: org.apache.hadoop.hbase.client.ScannerTimeoutException: 80389ms passed since the last invocation, timeout is currently set to 60000
while executing SQL: "SELECT ((( EXTRACT(YEAR FROM "POS_DTL"."DOC_DATE") * 10000) + ( EXTRACT(MONTH FROM 
{code}

Actually this timeout setting is the hbase scanner timeout "hbase.client.scanner.timeout.period", whose default value is also 60 seconds; Kylin didn't overwrite this, so find and change hbase-site.xml will customize this;

The three parameters in HadoopUtil.newHBaseConfiguration() is to avoid Kylin be hung on HBase, and reduce the retries times so that user can get error when HBase is in problem; So I will mark this problem as "Won't fix".

> Make hbase client configurations like timeout configurable
> ----------------------------------------------------------
>
>                 Key: KYLIN-1013
>                 URL: https://issues.apache.org/jira/browse/KYLIN-1013
>             Project: Kylin
>          Issue Type: Improvement
>          Components: Storage - HBase
>    Affects Versions: v1.0, v0.7.2, v0.7.1
>            Reporter: Shaofeng SHI
>            Assignee: Shaofeng SHI
>             Fix For: v1.1
>
>
> User askes how to set the timeout on HBase connection; In his case, some query will report timeout error after 60 seconds, from their observation it need about 77 seconds to finish; They hope this timeout value can be customized;
> Checked the code, in org.apache.kylin.common.util.HadoopUtil.newHBaseConfiguration(), there are some hard-coded configurations:
> {code}
>         // reduce rpc retry
>         conf.set(HConstants.HBASE_CLIENT_PAUSE, "3000");
>         conf.set(HConstants.HBASE_CLIENT_RETRIES_NUMBER, "5");
>         conf.set(HConstants.HBASE_CLIENT_OPERATION_TIMEOUT, "60000");
> {code}



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