[
https://issues.apache.org/jira/browse/HDFS-8696?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14726293#comment-14726293
]
Bob Hansen commented on HDFS-8696:
----------------------------------
With the patched hdfs, we're still seeing slowdowns.
For the 1-byte reads:
99.7% of requests have an average latency of 4.5ms
0.3% of requests have an average latency of 870 ms (selected for latency >
500ms)
So we're spending 34% of our wall-clock time waiting for the slow reads. This
is better than the unpatched 2.2.6, but still a bit rough.
Running the same experiment against 2.2.0.0 gave 0 reads at more than 500ms.
> Reduce the variances of latency of WebHDFS
> ------------------------------------------
>
> Key: HDFS-8696
> URL: https://issues.apache.org/jira/browse/HDFS-8696
> Project: Hadoop HDFS
> Issue Type: Improvement
> Components: webhdfs
> Affects Versions: 2.7.0
> Reporter: Xiaobing Zhou
> Assignee: Xiaobing Zhou
> Attachments: HDFS-8696.1.patch, HDFS-8696.2.patch, HDFS-8696.3.patch
>
>
> There is an issue that appears related to the webhdfs server. When making two
> concurrent requests, the DN will sometimes pause for extended periods (I've
> seen 1-300 seconds), killing performance and dropping connections.
> To reproduce:
> 1. set up a HDFS cluster
> 2. Upload a large file (I was using 10GB). Perform 1-byte reads, writing
> the time out to /tmp/times.txt
> {noformat}
> i=1
> while (true); do
> echo $i
> let i++
> /usr/bin/time -f %e -o /tmp/times.txt -a curl -s -L -o /dev/null
> "http://<namenode>:50070/webhdfs/v1/tmp/bigfile?op=OPEN&user.name=root&length=1";
> done
> {noformat}
> 3. Watch for 1-byte requests that take more than one second:
> tail -F /tmp/times.txt | grep -E "^[^0]"
> 4. After it has had a chance to warm up, start doing large transfers from
> another shell:
> {noformat}
> i=1
> while (true); do
> echo $i
> let i++
> (/usr/bin/time -f %e curl -s -L -o /dev/null
> "http://<namenode>:50070/webhdfs/v1/tmp/bigfile?op=OPEN&user.name=root");
> done
> {noformat}
> It's easy to find after a minute or two that small reads will sometimes
> pause for 1-300 seconds. In some extreme cases, it appears that the
> transfers timeout and the DN drops the connection.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
