[jira] [Resolved] (KUDU-2270) Allow long RPC duration which triggers logging to be configurable

["Michael Ho (JIRA)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/KUDU-2270?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Michael Ho resolved KUDU-2270.
------------------------------
       Resolution: Fixed
    Fix Version/s: 1.7.0

Fixed at commit https://github.com/apache/kudu/commit/656ed6a844174d4faacd64702f16c6d9948f3eee

KUDU-2270: Add a flag to control logging in RpczStore::LogTrace()

This change adds a new flag FLAGS_rpc_duration_too_long_ms
which controls the duration above which a RPC is considered
too long and is logged at INFO level in the log. Previously,
this threshold is hardcoded to 1000ms which may be too short
for a busy Impalad demon, leading to massive log spew.

Change-Id: Ie587ee602e83bb65d74f7ee622a9bc47897f2574
Reviewed-on: http://gerrit.cloudera.org:8080/9117
Reviewed-by: Todd Lipcon <to...@apache.org>
Tested-by: Kudu Jenkins

> Allow long RPC duration which triggers logging to be configurable
> -----------------------------------------------------------------
>
>                 Key: KUDU-2270
>                 URL: https://issues.apache.org/jira/browse/KUDU-2270
>             Project: Kudu
>          Issue Type: Improvement
>          Components: rpc
>    Affects Versions: 1.7.0
>            Reporter: Michael Ho
>            Assignee: Michael Ho
>            Priority: Minor
>             Fix For: 1.7.0
>
>
> {{Rpczstore::LogTrace}} will currently log some details about a RPC when the duration between when the RPC is accepted and when its RPC handler completes takes longer than 1000 ms. While this threshold may be reasonable for Kudu, a busy Impala demon utilizing KRPC can easily exceed this threshold. It'd be nice to allow this threshold of 1000ms to be configurable to reduce the log spew when running Impalad with KRPC. Please also see IMPALA-6356.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)