[
https://issues.apache.org/jira/browse/HDFS-14179?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16747297#comment-16747297
]
Íñigo Goiri commented on HDFS-14179:
------------------------------------
The main overhead would be the two new {{System.currentTimeMillis()}} calls,
right?
The log messages are guarded so it should be fairly fast.
[~shubham.dewan] can you take some quick performance numbers for the approach?
> BlockReaderRemote#readNextPacket() should log the waiting time for packet
> read.
> -------------------------------------------------------------------------------
>
> Key: HDFS-14179
> URL: https://issues.apache.org/jira/browse/HDFS-14179
> Project: Hadoop HDFS
> Issue Type: Bug
> Components: hdfs-client
> Affects Versions: 3.0.0
> Reporter: Surendra Singh Lilhore
> Assignee: Shubham Dewan
> Priority: Major
> Labels: newbie
> Attachments: HDFS-14179.001.patch, HDFS-14179.002.patch,
> HDFS-14179.003.patch
>
>
> Sometime read is reported very slow due to disk or some other reason.
> {{BlockReaderRemote#readNextPacket()}} should print the datanode IP and
> waiting time in trace log.
> {code:java}
> //Read packet headers.
> packetReceiver.receiveNextPacket(in);
> PacketHeader curHeader = packetReceiver.getHeader();
> curDataSlice = packetReceiver.getDataSlice();
> assert curDataSlice.capacity() == curHeader.getDataLen();
> LOG.trace("DFSClient readNextPacket got header {}", curHeader);{code}
>
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: hdfs-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: hdfs-issues-h...@hadoop.apache.org
