hadoop-hdfs-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Zhe Zhang (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HDFS-9918) Erasure Coding: Sort located striped blocks based on decommissioned states
Date Wed, 30 Mar 2016 22:07:25 GMT

    [ https://issues.apache.org/jira/browse/HDFS-9918?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15218965#comment-15218965
] 

Zhe Zhang commented on HDFS-9918:
---------------------------------

In my original comment when I said preserving index ordering, I meant something like below
(using Rakesh's example):
bq. Assume in the below list 0, 1, 7, 8 is decommissioned and extra redundant index is 1.
bq. 0, 1', 2, 3, 4, 5, 6, 7', 8', 0', 1, 7, 8, 1
In the above, I marked decommissioning internal blocks with a {{'}}. I think we should order
the list as:
{code}
0, 1, 2, 3, 4, 5, 6, 7, 8, 1, 0', 1', 7', 8'
{code}

The main motivation is to enable future optimization HDFS-8032. So actually I think we should
do a decomm-based sorting within a block index, and only "swap" 2 internal blocks with the
same block index.

In the v8 patch, I think we can use a {{DECOM_COMPARATOR}}-based sort on each list in {{blkIndex2LocationsMap}}.
Then for the first {{numDataUnits}} slots in the result, we take the first location from the
index. From sorting priority's perspective, I think a decommissioning internal block is still
more useful then a redundant internal block.

LMK if you guys agree. 

> Erasure Coding: Sort located striped blocks based on decommissioned states
> --------------------------------------------------------------------------
>
>                 Key: HDFS-9918
>                 URL: https://issues.apache.org/jira/browse/HDFS-9918
>             Project: Hadoop HDFS
>          Issue Type: Sub-task
>            Reporter: Rakesh R
>            Assignee: Rakesh R
>         Attachments: HDFS-9918-001.patch, HDFS-9918-002.patch, HDFS-9918-003.patch, HDFS-9918-004.patch,
HDFS-9918-005.patch, HDFS-9918-006.patch, HDFS-9918-007.patch, HDFS-9918-008.patch
>
>
> This jira is a follow-on work of HDFS-8786, where we do decommissioning of datanodes
having striped blocks.
> Now, after decommissioning it requires to change the ordering of the storage list so
that the decommissioned datanodes should only be last node in list.
> For example, assume we have a block group with storage list:-
> d0, d1, d2, d3, d4, d5, d6, d7, d8, d9
> mapping to indices
> 0, 1, 2, 3, 4, 5, 6, 7, 8, 2
> Here the internal block b2 is duplicated, locating in d2 and d9. If d2 is a decommissioning
node then should switch d2 and d9 in the storage list.
> Thanks [~jingzhao] for the [discussions|https://issues.apache.org/jira/browse/HDFS-8786?focusedCommentId=15180415&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-15180415]



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

Mime
View raw message