[
https://issues.apache.org/jira/browse/HDFS-6982?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14208879#comment-14208879
]
Andrew Wang commented on HDFS-6982:
-----------------------------------
bq. The configuration keys are already there. I have some constants in TopConf
though. Did you mean to move those top-specific constants to DFSConfigKeys?
Right now there are some prefixes that are concatenated to generate the full
string. Since there are only 3 config keys right now, I think it'd be nicer to
just write them out, i.e. "dfs.namenode.top.window.buckets".
bq. We actually use jmx for plotting the data exported by nntop, and that is
why one reporting period is sufficient. For html view, the web component
directly contacts the TopMetrics and retrieves the top users for all reporting
periods.
I guess this is reasonable, though it does differ a bit from how the new webUI
works. [~wheat9] do you have any feelings here?
bq. <counters never go back to 0>
Could you file a follow-on JIRA for this issue, and post the patch? It's fine
to do it before or after, but it seems like we definitely need it in the end
state either way.
Thanks Maysam!
> nntop: top-like tool for name node users
> -----------------------------------------
>
> Key: HDFS-6982
> URL: https://issues.apache.org/jira/browse/HDFS-6982
> Project: Hadoop HDFS
> Issue Type: New Feature
> Reporter: Maysam Yabandeh
> Assignee: Maysam Yabandeh
> Attachments: HDFS-6982.patch, HDFS-6982.v2.patch, HDFS-6982.v3.patch,
> HDFS-6982.v4.patch, HDFS-6982.v5.patch, HDFS-6982.v6.patch,
> nntop-design-v1.pdf
>
>
> In this jira we motivate the need for nntop, a tool that, similarly to what
> top does in Linux, gives the list of top users of the HDFS name node and
> gives insight about which users are sending majority of each traffic type to
> the name node. This information turns out to be the most critical when the
> name node is under pressure and the HDFS admin needs to know which user is
> hammering the name node and with what kind of requests. Here we present the
> design of nntop which has been in production at Twitter in the past 10
> months. nntop proved to have low cpu overhead (< 2% in a cluster of 4K
> nodes), low memory footprint (less than a few MB), and quite efficient for
> the write path (only two hash lookup for updating a metric).
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
