[
https://issues.apache.org/jira/browse/HDDS-699?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16754403#comment-16754403
]
Tsz Wo Nicholas Sze commented on HDDS-699:
------------------------------------------
Thanks [~Sammi] for working the patch. Some comments/questions
- Do you expect NetConf be set by users/admins? If not, let's rename it
something like NetConstants. In hadoop, conf is supposed to be set by
users/admin.
- NetworkTopology uses the single netlock approach for the entire data
structure. It has been a performance bottleneck in HDFS for a long time. I
wonder if we could make Node and InnerNode threadsafe:
-* the NodeImpl can be immutable so that the accesses do not need any lock.
-* childrenMap in InnerNodeImpl can be changed to ConcurrentHashMap
-* Do not maintain numOfLeaves in Root. For the other InnerNode, numOfLeaves
is protected by a lock in getNumOfLeaves(), add(..) and remove(..)
- Remove NetworkTopology.random since it has a race condition. Use
ThreadLocalRandom.
> Detect Ozone Network topology
> -----------------------------
>
> Key: HDDS-699
> URL: https://issues.apache.org/jira/browse/HDDS-699
> Project: Hadoop Distributed Data Store
> Issue Type: Sub-task
> Reporter: Xiaoyu Yao
> Assignee: Sammi Chen
> Priority: Major
> Attachments: HDDS-699.00.patch, HDDS-699.01.patch
>
>
> Traditionally this has been implemented in Hadoop via script or customizable
> java class. One thing we want to add here is the flexible multi-level support
> instead of fixed levels like DC/Rack/NG/Node.
--
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
