[
https://issues.apache.org/jira/browse/HDFS-8650?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15012698#comment-15012698
]
Walter Su commented on HDFS-8650:
---------------------------------
bq. as soon as an block group been written to DN, we reuse the 9
StripedDataStreamer to write the next block group, right?
You are right.
1. When we have a failed streamer, we will replace it when we start write to
next block group.
2. HDFS-9079 considers to use 9 new streamers for every block group.
The above 2 circumstances still won't create too many threads.
And thank you very much.
> Erasure Coding: use thread pool for StripedDataStreamer
> -------------------------------------------------------
>
> Key: HDFS-8650
> URL: https://issues.apache.org/jira/browse/HDFS-8650
> Project: Hadoop HDFS
> Issue Type: Sub-task
> Reporter: Walter Su
> Assignee: GAO Rui
> Priority: Minor
> Attachments: HDFS-8650-trunk.01.patch
>
>
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
