[
https://issues.apache.org/jira/browse/HDFS-12696?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16235752#comment-16235752
]
Nanda kumar commented on HDFS-12696:
------------------------------------
*Code flow*
{noformat}
DataNode#main
secureMain
createDataNode
instantinateDataNode --> This starts all BPOfferServices in
blockPoolManager (below flow) - blockPoolManager.startAll is called
runDatanodeDaemon
blockPoolManager.startAll() --> Calling it again here
instantinateDataNode
makeInstance
DataNode.startDatanode
blockPoolManager = new BlockPoolManager()
blockPoolManager.refreshNamenodes
blockPoolManager.doRefreshNamenodes
blockPoolManager.createBPOS
BPOfferService#Constructor
blockPoolManager.startAll() --> Calling startAll for the first time
{noformat}
> BlockPoolManager#startAll is called twice during DataNode startup
> -----------------------------------------------------------------
>
> Key: HDFS-12696
> URL: https://issues.apache.org/jira/browse/HDFS-12696
> Project: Hadoop HDFS
> Issue Type: Bug
> Components: datanode
> Reporter: Nanda kumar
> Assignee: Nanda kumar
> Priority: Minor
> Attachments: HDFS-12696.000.patch
>
>
> As part of Datanode startup, {{BlockPoolManager#startAll}} which starts all
> {{BPServiceActor}} threads is called twice.
> First in {{Datanode}} constructor, {{Datanode#startDataNode}} is called which
> does {{BlockPoolManager#refreshNamenodes}} inside which we do {{startAll}}
> And as part of {{Datanode#runDatanodeDaemon}} we again call
> {{BlockPoolManager#startAll}}.
> Since {{BPServiceActor}} checks if {{bpThread}} is already running, before
> starting them again, the second call is ignored.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
---------------------------------------------------------------------
To unsubscribe, e-mail: hdfs-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: hdfs-issues-h...@hadoop.apache.org
