[
https://issues.apache.org/jira/browse/HDFS-11060?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15664815#comment-15664815
]
Ravi Prakash commented on HDFS-11060:
-------------------------------------
Hi Lantao! Could you please go through
https://wiki.apache.org/hadoop/HowToContribute ?
Ideally we'd like to see a patch file uploaded to the JIRA. HadoopQA would then
run it through some pre-commit tests and give a +1 (or -1)
>From the pull request I see that you have simply made the 100 configurable.
>However for pagination to work, we'd have to provide another API that takes an
>*offset/pageid* along with the {{maxCorruptFileBlocksReturn}} , won't we?
>https://github.com/apache/hadoop/blob/ff0b99eafeda035ebe0dc82cfe689808047a8893/hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/namenode/NameNodeRpcServer.java#L1281
> .
> make DEFAULT_MAX_CORRUPT_FILEBLOCKS_RETURNED configurable
> ---------------------------------------------------------
>
> Key: HDFS-11060
> URL: https://issues.apache.org/jira/browse/HDFS-11060
> Project: Hadoop HDFS
> Issue Type: Improvement
> Components: hdfs
> Reporter: Lantao Jin
> Priority: Minor
>
> Current, the easiest way to determine which blocks is missing is using NN web
> UI or JMX. Unfortunately, because the
> DEFAULT_MAX_CORRUPT_FILEBLOCKS_RETURNED=100 is hard code in FSNamesystem,
> only 100 missing blocks can be return by UI and JMX. Even the result of URL
> "https://nn:50070/fsck?listcorruptfileblocks=1&path=%2F"; is limited by this
> hard code value too.
> I did know FSCK can return more than 100 result but due to the security
> reason(with kerberos), it is very hard to involve to costumer programs and
> scripts.
> So I think it should add a configurable var "maxCorruptFileBlocksReturned" to
> fix above case.
> If community also think it's worth to do, I will patch this. If not, please
> feel free to tell me the reason.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
