[
https://issues.apache.org/jira/browse/HDFS-13840?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16604418#comment-16604418
]
Brahma Reddy Battula edited comment on HDFS-13840 at 9/5/18 1:48 PM:
---------------------------------------------------------------------
{quote}1. In {{checkReplicaCorrupt()}}, {{isStriped()}} block check it is not
required. EC file also handle gstamp same as continues file.
{quote}
Done. I was misunderstand initially, I was thinking stripped testcases are
expected with GS 0 .Hence I didn't modified it.
{quote}2. Why the below change is required ?, if block is marked as corrupted
and it is corrupted during write then it will marked as invalid.
{quote}
Since I am adding to corrupt storages which will be updated with corrupt.So
when same DN report with new GS,we need to overwrite it. I added the comment
for same.
was (Author: brahmareddy):
bq.1. In {{checkReplicaCorrupt()}}, {{isStriped()}} block check it is not
required. EC file also handle gstamp same as continues file.
Done. I was misunderstand initially, I was thinking stripped testcases are
expected with GS 0 .Hence I didn't modified it.
bq.2. Why the below change is required ?, if block is marked as corrupted and
it is corrupted during write then it will marked as invalid.
Since I am adding to corrupt storages will be updated with corrupt.So when same
DN report with new GS,we need to overwrite it. I added the comment for same.
> RBW Blocks which are having less GS should be added to Corrupt
> --------------------------------------------------------------
>
> Key: HDFS-13840
> URL: https://issues.apache.org/jira/browse/HDFS-13840
> Project: Hadoop HDFS
> Issue Type: Bug
> Reporter: Brahma Reddy Battula
> Assignee: Brahma Reddy Battula
> Priority: Minor
> Attachments: HDFS-13840-002.patch, HDFS-13840-003.patch,
> HDFS-13840-004.patch, HDFS-13840-005.patch, HDFS-13840.patch
>
>
> # Start two DN's (DN1,DN2).
> # Write fileA with rep=2 ( dn't close)
> # Stop DN1.
> # Write some data to fileA.
> # restart the DN1
> # Get the blocklocations of fileA.
> Here RWR state block will be reported on DN restart and added to locations.
> IMO,RWR blocks which having less GS shouldn't added, as they give false
> postive (anyway read can be failed as it's genstamp is less)
--
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
