[
https://issues.apache.org/jira/browse/HDFS-4417?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13556500#comment-13556500
]
Colin Patrick McCabe commented on HDFS-4417:
--------------------------------------------
This is a very good find, thanks guys.
I have to think about this a little bit more. I would like to avoid splitting
{{DomainSocketFactory#create}} into two functions if it is at all possible. I
feel like we could be a little bit smarter in PeerCache and avoid a lot of
these problems. And yeah, we need a unit test. Mind if I take this one?
If we do have a mismatch between the domain socket keepalive and the length of
time we cache sockets in PeerCache, we obviously should fix that-- trying to
use sockets that we ought to know are stale is not smart. (Obviously, there
will always be some mismatches-- if the server's keepalive changes and not all
clients are updated, etc.)
> HDFS-347: fix case where local reads get disabled incorrectly
> -------------------------------------------------------------
>
> Key: HDFS-4417
> URL: https://issues.apache.org/jira/browse/HDFS-4417
> Project: Hadoop HDFS
> Issue Type: Sub-task
> Components: datanode, hdfs-client, performance
> Reporter: Todd Lipcon
> Assignee: Todd Lipcon
> Attachments: hdfs-4417.txt
>
>
> In testing HDFS-347 against HBase (thanks [~jdcryans]) we ran into the
> following case:
> - a workload is running which puts a bunch of local sockets in the PeerCache
> - the workload abates for a while, causing the sockets to go "stale" (ie the
> DN side disconnects after the keepalive timeout)
> - the workload starts again
> In this case, the local socket retrieved from the cache failed the
> newBlockReader call, and it incorrectly disabled local sockets on that host.
> This is similar to an earlier bug HDFS-3376, but not quite the same.
> The next issue we ran into is that, once this happened, it never tried local
> sockets again, because the cache held lots of TCP sockets. Since we always
> managed to get a cached socket to the local node, it didn't bother trying
> local read again.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
