[
https://issues.apache.org/jira/browse/HDFS-15650?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17224882#comment-17224882
]
Yushi Hayasaka commented on HDFS-15650:
---------------------------------------
[~weichiu] Yes, but I think we need to use the same value between client and DN
because it is possible to happen the scenario below:
* If the client is set `dfs.client.socket-timeout` to 3000 and the DN uses the
default value (60000), the client disconnects the connection despite the DN
tries to fetch checksums and it continues the process even if no one want.
I think it may exhaust the handler thread if there is a lot of request like
above... WDYT?
> Make the socket timeout for computing checksum of striped blocks configurable
> -----------------------------------------------------------------------------
>
> Key: HDFS-15650
> URL: https://issues.apache.org/jira/browse/HDFS-15650
> Project: Hadoop HDFS
> Issue Type: Improvement
> Components: datanode, ec, erasure-coding
> Reporter: Yushi Hayasaka
> Priority: Minor
> Labels: pull-request-available
> Time Spent: 20m
> Remaining Estimate: 0h
>
> Regarding the DataNode tries to get the checksum of EC internal blocks from
> another DataNode for computing the checksum of striped blocks, the timeout is
> hard-coded now, but it should be configurable.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
