[
https://issues.apache.org/jira/browse/HDFS-14169?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16728296#comment-16728296
]
Ayush Saxena commented on HDFS-14169:
-------------------------------------
Thanx [~surendrasingh] for the pointers at HDFS-14167.
Just on a doubt.
If we didn't get it.we are say sending back number of stale datanodes to be
0.Which is valid.Will it not develop a wrong perspective? The person fetching
it will consider it to be true and won't dig up further. Always believed send
it correct or don't send it. And How come a Disclaimer is a valid result?
Well as pointed,If it is having some Technical drawbacks.I would too prefer
pulling this down. :)
{quote}No need, metrics is count for system status. It should not return -1
{quote}
[~surendrasingh] Should we replace all -1's in the class with 0(saw couple of
them). As quoted default is 0 .I guess one standard should be followed. And If
0 is the one being followed we should stick to it.
Somewhere -1 Somewhere 0.It's just confusing.We should consider making it
little consistent.When we don't have it we send the default 0.
[~elgoiri] I guess you in HDFS-12335 had set up this approach of having -1.Can
you help how at that time the technical difficulties were taken care and the
default value of 0 wasn't used? What was the idea over there?
> RBF: Correct the returned value in case of IOException in
> NamenodeBeanMetrics#getFederationMetrics
> --------------------------------------------------------------------------------------------------
>
> Key: HDFS-14169
> URL: https://issues.apache.org/jira/browse/HDFS-14169
> Project: Hadoop HDFS
> Issue Type: Sub-task
> Reporter: Ayush Saxena
> Assignee: Ayush Saxena
> Priority: Major
> Attachments: HDFS-14169-HDFS-13891-01.patch
>
>
> Presently in case of IOException the the metrics value returned is 0 which is
> a legal entry.Better to change to a value which could indicate that the value
> hasn't been actually fetched.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
