[
https://issues.apache.org/jira/browse/HDDS-1511?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16836659#comment-16836659
]
Arpit Agarwal commented on HDDS-1511:
-------------------------------------
bq. Perhaps a better investment is to write code that handles the container
creation or chunk write failure, or come up with an algorithm to look at what
HDFS is doing too.
[~anu], you are absolutely right. We should and we will make sure we robustly
handle IO failures.
This particular Jira fixes a very limited scenario of putting new containers on
obviously full volumes. It is not a substitute for robust failure handling.
> Space tracking for Open Containers in HDDS Volumes
> --------------------------------------------------
>
> Key: HDDS-1511
> URL: https://issues.apache.org/jira/browse/HDDS-1511
> Project: Hadoop Distributed Data Store
> Issue Type: Improvement
> Components: Ozone Datanode
> Reporter: Supratim Deka
> Assignee: Supratim Deka
> Priority: Major
> Attachments: HDDS-1511.000.patch
>
>
> For every HDDS Volume, track the space usage in open containers. Introduce a
> counter committedBytes in HddsVolume - this counts the remaining space in
> Open containers until they reach max capacity. The counter is incremented (by
> container max capacity) for every container create. And decremented (by chunk
> size) for every chunk write.
> Space tracking for open containers will enable adding a safety check during
> container create.
> If there is not sufficient free space in the volume, the container create
> operation can be failed.
> The scope of this jira is to just add the space tracking for Open Containers.
> Checking for space and failing container create will be introduced in a
> subsequent jira.
--
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
