[
https://issues.apache.org/jira/browse/HDDS-6460?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Stephen O'Donnell updated HDDS-6460:
------------------------------------
Fix Version/s: EC-Branch
Resolution: Fixed
Status: Resolved (was: Patch Available)
> EC: Adjust requested size by EC DataNum in WritableECContainerProvider
> ----------------------------------------------------------------------
>
> Key: HDDS-6460
> URL: https://issues.apache.org/jira/browse/HDDS-6460
> Project: Apache Ozone
> Issue Type: Sub-task
> Components: SCM
> Reporter: Stephen O'Donnell
> Assignee: Stephen O'Donnell
> Priority: Major
> Labels: pull-request-available
> Fix For: EC-Branch
>
>
> In HDDS-6384 we decided that the size tracked in SCM of an EC container is
> the usedBytes in the first replica or any parity replica. In
> WritableECContainerProvider, we currently check if there is enough space in
> the container by comparing the container UsedBytes against the blockSize
> requested. However the blockSize needs to be divided by the EC DataNum, as it
> will be stripped across blocks in the group. There was a TODO left in the
> code to address this decision later:
> {code}
> ContainerInfo containerInfo = getContainerFromPipeline(pipeline);
> // TODO - For EC, what is the block size? If the client says 128MB,
> // is that 128MB / 6 (for EC-6-3?)
> if (containerInfo == null
> || !containerHasSpace(containerInfo, size)) {
> // This is O(n), which isn't great if there are a lot of pipelines
> // and we keep finding pipelines without enough space.
> existingPipelines.remove(pipeline);
> pipelineManager.closePipeline(pipeline, true);
> {code}
> After the decision in HDDS-6384, we should remove this TODO and divide the
> requested size by the dataNum to get the approx space needed in the first
> replica to hold the new block.
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
