[
https://issues.apache.org/jira/browse/HDFS-7112?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14162629#comment-14162629
]
Chris Nauroth commented on HDFS-7112:
-------------------------------------
bq. For asyncLazyPersistService, we will only call
asyncLazyPersistService#addVolume when the storage type of the volume is
RAM_DISK.
I see the volume will get added to {{asyncLazyPersistService}} if
{{!storageType.isTransient()}}. This would mean every "traditional" volume
would get added, and therefore a thread would be allocated. I think Arpit was
pointing out that for existing deployments not using {{RAM_DISK}}, this would
trigger allocation of additional unneeded threads. Ideally, the non-transient
volumes would only get added to {{asyncLazyPersistService}} if there is at
least one {{RAM_DISK}} volume configured. This way, the extra threads wouldn't
be allocated in deployments that don't use any {{RAM_DISK}} volumes.
Aside from that, the patch looks good. It appears that all other prior
feedback from Arpit has been addressed.
> LazyWriter should use either async IO or one thread per physical disk
> ---------------------------------------------------------------------
>
> Key: HDFS-7112
> URL: https://issues.apache.org/jira/browse/HDFS-7112
> Project: Hadoop HDFS
> Issue Type: Sub-task
> Components: datanode
> Affects Versions: HDFS-6581
> Reporter: Arpit Agarwal
> Assignee: Xiaoyu Yao
> Fix For: 2.6.0
>
> Attachments: HDFS-7112.0.patch, HDFS-7112.1.patch, HDFS-7112.2.patch
>
>
> The LazyWriter currently uses synchronous IO and a single thread. This limits
> the throughput to that of a single disk. Using either async overlapped IO or
> one thread per physical disk will improve the write throughput.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
