You are viewing a plain text version of this content. The canonical link for it is here.
Posted to hdfs-dev@hadoop.apache.org by "Supratim Deka (JIRA)" <ji...@apache.org> on 2019/05/20 03:31:00 UTC

[jira] [Created] (HDDS-1559) Include committedBytes to determine Out of Space in VolumeChoosingPolicy

Supratim Deka created HDDS-1559:
-----------------------------------

             Summary: Include committedBytes to determine Out of Space in VolumeChoosingPolicy
                 Key: HDDS-1559
                 URL: https://issues.apache.org/jira/browse/HDDS-1559
             Project: Hadoop Distributed Data Store
          Issue Type: Improvement
          Components: Ozone Datanode
            Reporter: Supratim Deka
            Assignee: Supratim Deka


This is a follow-up from HDDS-1511 and HDDS-1535

Currently  when creating a new Container, the DN invokes RoundRobinVolumeChoosingPolicy:chooseVolume(). This routine checks for (volume available space > container max size). If no eligible volume is found, the policy throws a DiskOutOfSpaceException. This is the current behaviour.

However, the computation of available space does not take into consideration the space
that is going to be consumed by writes to existing containers which are still Open and accepting chunk writes.

This Jira proposes to enhance the space availability check in chooseVolume by inclusion of committed space(committedBytes in HddsVolume) in the equation.

The handling/management of the exception in Ratis will not be modified in this Jira. That will be scoped separately as part of Datanode IO Failure handling work.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

---------------------------------------------------------------------
To unsubscribe, e-mail: hdfs-dev-unsubscribe@hadoop.apache.org
For additional commands, e-mail: hdfs-dev-help@hadoop.apache.org