hadoop-hdfs-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Senthilkumar (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HDFS-10977) Balancer should query NameNode with a timeout
Date Fri, 07 Oct 2016 04:59:20 GMT

    [ https://issues.apache.org/jira/browse/HDFS-10977?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15554136#comment-15554136
] 

Senthilkumar commented on HDFS-10977:
-------------------------------------

[~zhz] ,  When i was working with balancer last month i too faced this issue ,

2016-09-08 06:32:06,574 DEBUG org.apache.hadoop.ipc.Client: IPC Client (685788708) connection
to host/10.103.108. 
201:8020 from hadoop/host@DOMAIN sending #49230 
2016-09-08 06:32:06,574 DEBUG org.apache.hadoop.ipc.ProtobufRpcEngine: Call: getBlocks took
0ms 
2016-09-08 06:32:06,575 DEBUG org.apache.hadoop.ipc.Client: IPC Client (685788708) connection
to host/10.103.108. 
201:8020 from hadoop/host@DOMAIN sending #49231 
2016-09-08 06:32:06,575 DEBUG org.apache.hadoop.ipc.Client: IPC Client (685788708) connection
to host/10.103.108. 
201:8020 from hadoop/host@DOMAIN got value #49229 
2016-09-08 06:32:06,575 DEBUG org.apache.hadoop.ipc.ProtobufRpcEngine: Call: getBlocks took
1ms 
2016-09-08 06:32:06,575 DEBUG org.apache.hadoop.ipc.Client: IPC Client (685788708) connection
to host/10.103.108. 
201:8020 from hadoop/host@DOMAIN got value #49230 
2016-09-08 06:32:06,575 DEBUG org.apache.hadoop.ipc.Client: IPC Client (685788708) connection
to host/10.103.108. 
201:8020 from hadoop/host@DOMAIN sending #49232 
2016-09-08 06:32:06,575 DEBUG org.apache.hadoop.ipc.ProtobufRpcEngine: Call: getBlocks took
1ms 
2016-09-08 06:32:06,575 DEBUG org.apache.hadoop.ipc.Client: IPC Client (685788708) connection
to host/10.103.108. 
201:8020 from hadoop/host@DOMAIN sending #49233 
2016-09-08 06:32:06,575 DEBUG org.apache.hadoop.ipc.Client: IPC Client (685788708) connection
to host/10.103.108. 
201:8020 from hadoop/host@DOMAIN got value #49231 
2016-09-08 06:32:06,575 DEBUG org.apache.hadoop.ipc.ProtobufRpcEngine: Call: getBlocks took
0ms 
2016-09-08 06:32:06,575 DEBUG org.apache.hadoop.ipc.Client: IPC Client (685788708) connection
to host/10.103.108. 
201:8020 from hadoop/host@DOMAIN sending #49234 
2016-09-08 06:32:06,575 DEBUG org.apache.hadoop.ipc.Client: IPC Client (685788708) connection
to host/10.103.108. 
201:8020 from hadoop/host@DOMAIN got value #49232 
2016-09-08 06:32:06,575 DEBUG org.apache.hadoop.ipc.ProtobufRpcEngine: Call: getBlocks took
0ms 
2016-09-08 06:32:06,575 DEBUG org.apache.hadoop.ipc.Client: IPC Client (685788708) connection
to host/10.103.108. 
201:8020 from hadoop/host@DOMAIN sending #49235 
2016-09-08 06:32:06,575 DEBUG org.apache.hadoop.ipc.Client: IPC Client (685788708) connection
to host/10.103.108. 
201:8020 from hadoop/host@DOMAIN got value #49233 
2016-09-08 06:32:06,575 DEBUG org.apache.hadoop.ipc.ProtobufRpcEngine: Call: getBlocks took
0ms 
2016-09-08 06:32:06,575 DEBUG org.apache.hadoop.ipc.Client: IPC Client (685788708) connection
to host/10.103.108. 
201:8020 from hadoop/host@DOMAIN got value #49234


Here is the TRACE where it took few hours and no blocks found .. I ended up in Restarting
balancer to make it work..


At last i started balancer with -include option ( by pulling LIVE DNs ) and it helped .. Not
sure including live nodes is the right solution ?? .. If it is the case i think problem is
balancer taking decomm/decommed nodes as well ??

Attaching the discussion which i started sometime back ..


> Balancer should query NameNode with a timeout
> ---------------------------------------------
>
>                 Key: HDFS-10977
>                 URL: https://issues.apache.org/jira/browse/HDFS-10977
>             Project: Hadoop HDFS
>          Issue Type: Bug
>          Components: balancer & mover
>            Reporter: Zhe Zhang
>            Assignee: Zhe Zhang
>         Attachments: HDFS-10977-reproduce.patch
>
>
> We found a case where {{Dispatcher}} was stuck at {{getBlockList}} *forever* (well, several
hours when we found it).



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

---------------------------------------------------------------------
To unsubscribe, e-mail: hdfs-issues-unsubscribe@hadoop.apache.org
For additional commands, e-mail: hdfs-issues-help@hadoop.apache.org


Mime
View raw message