hadoop-hdfs-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Chen Liang (JIRA)" <j...@apache.org>
Subject [jira] [Comment Edited] (HDFS-12000) Ozone: Container : Add key versioning support
Date Thu, 13 Jul 2017 23:24:00 GMT

    [ https://issues.apache.org/jira/browse/HDFS-12000?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16086573#comment-16086573
] 

Chen Liang edited comment on HDFS-12000 at 7/13/17 11:23 PM:
-------------------------------------------------------------

Thank [~anu] for updating the doc!!

One question though (I think I asked this some time earlier). 4.1 only mentioned range put.
I'm assuming seek() will be implemented here because the term streams is being used here.
So what happens when I, say, write X bytes to location 10, then do a seek() to another location,
say 100000, and write some Y bytes again? I guess this will be treated as two separate range
put?If so, does it imply that we should always commit before seek() or anything? I think we
talked about vectored IO here but I'm still not sure how will vectored IO help for this particular
case. Because these two writes (before seek() and after seek()) can be any of the situations:
disjoint, contiguous or overlap.

Another related question, do we allow seek to somewhere beyond current size of the key and
write? e.g. write a block of 1000 bytes, which covered 0 ~ 999 offset. Do we allow seek to
some offset beyond 999 and write, say, write to 2000 offset? Seems nothing prevents us from
doing it though.


was (Author: vagarychen):
Thank [~anu] for updating the doc!!

One question though (I think I asked this some time earlier). 4.1 only mentioned range put.
I'm assuming seek() will be implemented here because the term streams is being used here.
So what happens when I, say, write X bytes to location 10, then do a seek() to another location,
say 100000, and write some Y bytes again? I guess this will be treated as two separate range
put?If so, does it imply that we should always commit before seek() or anything? I think we
talked about vectored IO here but I'm still not sure how will vectored IO help for this particular
case. Because these two writes (before seek() and after seek()) can be any of the situations:
disjoint, contiguous or overlap.

> Ozone: Container : Add key versioning support
> ---------------------------------------------
>
>                 Key: HDFS-12000
>                 URL: https://issues.apache.org/jira/browse/HDFS-12000
>             Project: Hadoop HDFS
>          Issue Type: Sub-task
>          Components: ozone
>    Affects Versions: HDFS-7240
>            Reporter: Anu Engineer
>            Assignee: Chen Liang
>         Attachments: OzoneVersion.001.pdf
>
>
> The rest interface of ozone supports versioning of keys. This support comes from the
containers and how chunks are managed to support this feature. This JIRA tracks that feature.
Will post a detailed design doc so that we can talk about this feature.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)

---------------------------------------------------------------------
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