[
https://issues.apache.org/jira/browse/HADOOP-16266?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16832090#comment-16832090
]
Chao Sun commented on HADOOP-16266:
-----------------------------------
Overall looks good to me. Mostly nits:
- In {{updateMetrics}} we are calling with {{details.get}} with a time unit
which is the same as the one in {{details}}. This will unnecessarily call the
{{convert}} on time unit. It would be great if we can avoid this.
- Not sure if we should update the documentation for RPC detailed metrics as
they now do not include lock time and is different from RPC metrics.
- Question on {{RpcCall#isOpen}} - this is used to determine whether a call
has been processed or not, but is implemented using
{{connection.channel.isOpen()}}, which is a little confusing to me. Isn't a
channel being used to serve multiple RPC calls, and is only closed when you run
into some errors or timed out? how can it be used to indicate whether a call
has been processed or not?
- Since we are adding quantile metrics for RPC wait time, we may need to
document this in {{Metrics.md}} as well. The current way for doing this is kind
of tedious though I know :-/
> Add more fine-grained processing time metrics to the RPC layer
> --------------------------------------------------------------
>
> Key: HADOOP-16266
> URL: https://issues.apache.org/jira/browse/HADOOP-16266
> Project: Hadoop Common
> Issue Type: Improvement
> Components: ipc
> Reporter: Christopher Gregorian
> Assignee: Christopher Gregorian
> Priority: Minor
> Labels: rpc
> Attachments: HADOOP-16266.001.patch, HADOOP-16266.002.patch,
> HADOOP-16266.003.patch, HADOOP-16266.004.patch, HADOOP-16266.005.patch,
> HADOOP-16266.006.patch, HADOOP-16266.007.patch, HADOOP-16266.008.patch,
> HADOOP-16266.009.patch, HADOOP-16266.010.patch
>
>
> Splitting off of HDFS-14403 to track the first part: introduces more
> fine-grained measuring of how a call's processing time is split up.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
