[
https://issues.apache.org/jira/browse/HDFS-8855?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14694104#comment-14694104
]
Xiaobing Zhou commented on HDFS-8855:
-------------------------------------
[~bobhansen]
I was using the sequential work load for results above.
{noformat}
#!/bin/bash
# segment, op=OPEN and offset are added to url_base
count=${count:-1000000}
#echo $count
url_base=${url_base:-"http://c6401.ambari.apache.org:50070/webhdfs/v1/tmp/bigfile"}
#echo $url_base
read_size=${read_size:-1}
#echo $read_size
file_size=${file_size:-$[ 1024 * 1024 * 1024 ]}
#echo $file_size
#namenode=${namenode:-`echo $url_base | grep -Po "(?<=http://)[^:/]*"`}
namenode="c6401.ambari.apache.org"
#echo $namenode
for i in `seq 1 $count` ; do
rand=$(od -N 4 -t uL -An /dev/urandom | tr -d " ")
#echo $rand
offset=$[ ( $rand % (file_size / read_size) * read_size )]
#echo $offset
url=$url_base?op=OPEN\&offset=$offset\&length=$read_size
#echo $url
curl -L "$url" > url.blah 2>/dev/null
#curl -L "$url"
if (( $i % 100 == 0 )) ; then
# Display the time
echo -n "$i ";
date +%H:%M:%S.%N
# Count the connections on the NameNode
ssh vagrant@$namenode "file=/tmp/netstat.out ; netstat -a > \$file ; echo
-n 'ESTABLISHED: '; echo -n \`grep -c ESTABLISHED \$file\` ; echo -n '
TIME_WAIT: '; echo -n \`grep -c TIME_WAIT \$file\` ; echo -n ' CLOSE_WAIT: ';
grep -c CLOSE_WAIT \$file"&
fi
# sleep $delay
done
{noformat}
By running one load generator, I got up to 2200 connections, two to 3200.
Also ran the concurrent workload, there's up to 7000 connections. Making
file_size=1 and read_size=1 does not necessarily exacerbate it. I think it has
something to do with my cluster. Three nodes being local VMs. NN/DN/SNN is
equally deployed to 3 nodes.
Let's first try to work on the cache of org.apache.hadoop.ipc.connection to
have a test again.
> Webhdfs client leaks active NameNode connections
> ------------------------------------------------
>
> Key: HDFS-8855
> URL: https://issues.apache.org/jira/browse/HDFS-8855
> Project: Hadoop HDFS
> Issue Type: Bug
> Components: webhdfs
> Environment: HDP 2.2
> Reporter: Bob Hansen
> Assignee: Xiaobing Zhou
>
> The attached script simulates a process opening ~50 files via webhdfs and
> performing random reads. Note that there are at most 50 concurrent reads,
> and all webhdfs sessions are kept open. Each read is ~64k at a random
> position.
> The script periodically (once per second) shells into the NameNode and
> produces a summary of the socket states. For my test cluster with 5 nodes,
> it took ~30 seconds for the NameNode to have ~25000 active connections and
> fails.
> It appears that each request to the webhdfs client is opening a new
> connection to the NameNode and keeping it open after the request is complete.
> If the process continues to run, eventually (~30-60 seconds), all of the
> open connections are closed and the NameNode recovers.
> This smells like SoftReference reaping. Are we using SoftReferences in the
> webhdfs client to cache NameNode connections but never re-using them?
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
