[ 
https://issues.apache.org/jira/browse/HDFS-12000?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Chen Liang updated HDFS-12000:
------------------------------
    Attachment: HDFS-12000-HDFS-7240.001.patch

Versioning is going to be a rather complex change, a number of steps will be 
involved, each as one JIRA. Post initial patch for the first step. More 
specifically this patch tries to do two things:

1. add version field to KSM key metadata.
    - each version corresponds to a list of blocks, one version is represented 
by one {{KsmKeyLocationListVersioned}} object
   - a Key {{KsmKeyInfo}} includes a list of {{KsmKeyLocationListVersioned}} 
objects
   - when a new block is allocated to the key (which happens when the range 
write gets added later), a new version is generated by adding the new block to 
existing, latest version.

2. add commitBlock syntax
   - according to design doc, when a block is allocated, it should not be 
written to metadata instantly, but rather, client needs to make a further 
commit block call, then a new version is generated and made visible.

Some workaround are added to not completely bust the IO code path. For example, 
ChunkGroupOutputStream still works likes it currently does by assuming there is 
always only one version and always add the key to metadata on allocation. I've 
commented the places that are just temporary workaround, please ignore these 
places when reviewing the patch...

Any comments are welcome!

> 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: HDFS-12000-HDFS-7240.001.patch, 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-unsubscr...@hadoop.apache.org
For additional commands, e-mail: hdfs-issues-h...@hadoop.apache.org

Reply via email to