hadoop-common-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Masatake Iwasaki (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HADOOP-12672) RPC timeout should not override IPC ping interval
Date Sun, 13 Mar 2016 00:17:33 GMT

    [ https://issues.apache.org/jira/browse/HADOOP-12672?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15192023#comment-15192023

Masatake Iwasaki commented on HADOOP-12672:

The value of {{Client#getTimeout}} is used in 2 places in the current hadoop code base.

* DfsClientConf.hdfsTimeout
* NameNodeProxiesClient#createNonHAProxyWithClientProtocol (HDFS-4646)

{{DfsClientConf.hdfsTimeout}} affects the renewal time of LeaseRenewer. There is no problem
to fix {{Client#getTimeout}} to return rgiht timeout value.

There is a compatibility concern about {{NameNodeProxiesClient#createNonHAProxyWithClientProtocol}}
since HDFS-4646 depends on the value of {{Client#getTimeout}} when {{ipc.client.ping = false}}.

ipc.client.ping = false
ipc.ping.interval = 6000 (default)
ipc.client.rpc-timeout.ms = 0(default)

{{Client#getTimeout}} currently returns 6000 on the configuration above. If we fix to return
0 as timeout value, existing cluster which uses {{ipc.client.ping = false}} expecting that
the timeout of namenode proxy is set to ipc.ping.interval lose the timeout setting.

> RPC timeout should not override IPC ping interval
> -------------------------------------------------
>                 Key: HADOOP-12672
>                 URL: https://issues.apache.org/jira/browse/HADOOP-12672
>             Project: Hadoop Common
>          Issue Type: Improvement
>          Components: ipc
>    Affects Versions: 2.8.0, 2.7.3, 2.6.4
>            Reporter: Masatake Iwasaki
>            Assignee: Masatake Iwasaki
>             Fix For: 2.8.0, 2.7.3, 2.6.5
>         Attachments: HADOOP-12672.001.patch, HADOOP-12672.002.patch, HADOOP-12672.003.patch,
> Currently if the value of ipc.client.rpc-timeout.ms is greater than 0, the timeout overrides
the ipc.ping.interval and client will throw exception instead of sending ping when the interval
is passed. RPC timeout should work without effectively disabling IPC ping.

This message was sent by Atlassian JIRA

View raw message