[
https://issues.apache.org/jira/browse/HADOOP-12325?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14707467#comment-14707467
]
Xiaoyu Yao commented on HADOOP-12325:
-------------------------------------
[~anu], thanks for updating the patch. Some further comments:
bq. I could not find any place where we were using WritableRpcEngine for real,
hence I did not make that change.
I've attached a caller hierarchy from IntelliJ for WritableRpcEngine#call. I
think we should logSlowRpcCalls for both ProtobufRpcEngine#call and
WritableRpcEngine#call. The easy way to do this is to refactor logSlowRpcCalls
from ProtobufRpcEngine.java to Server.java. This way, both
ProtocoBufRpcEngine#call and WritableRpcEngine#call can simply call
server#logSlowRpcCalls instead.
There is an early JIRA HDFS-5180 with patches available sharing the similar
idea. We can resolve that one once HADOOP-12325 is fixed.
> RPC Metrics : Add the ability track and log slow RPCs
> -----------------------------------------------------
>
> Key: HADOOP-12325
> URL: https://issues.apache.org/jira/browse/HADOOP-12325
> Project: Hadoop Common
> Issue Type: Improvement
> Components: ipc, metrics
> Affects Versions: 2.7.1
> Reporter: Anu Engineer
> Assignee: Anu Engineer
> Attachments: HADOOP-12325.001.patch, HADOOP-12325.002.patch,
> HADOOP-12325.003.patch
>
>
> This JIRA proposes to add a counter called RpcSlowCalls and also a
> configuration setting that allows users to log really slow RPCs. Slow RPCs
> are RPCs that fall at 99th percentile. This is useful to troubleshoot why
> certain services like name node freezes under heavy load.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
