[
https://issues.apache.org/jira/browse/HDDS-14388?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18053458#comment-18053458
]
Siddhant Sangwan commented on HDDS-14388:
-----------------------------------------
The formula for calculating Non Ozone Used space is overviewStorageCard.tsx:
{code}
{
key: 'non-ozone-used',
usage: <Tag key='non-ozone-used' color='blue'>Non Ozone Used</Tag>,
size: size(storageReport.capacity - storageReport.remaining -
storageReport.used),
desc: 'Size of data used by Ozone for other files like logs, DB data etc.'
}
{code}
Note that `capacity` here is not the raw disk capacity, but actually raw disk
capacity - du.reserved. Similarly `remaining` is adjusted for du.reserved.
Meanwhile, `used` is the value obtained from running DU on the datanode data
directory, where the block files are located.
So if du.reserved is increased, non-ozone-used will decrease, which is the
behaviour seen in this jira.
Problems:
1. `non-ozone-used` is not the correct term for "Size of data used by Ozone for
other files like logs, DB data etc." Non ozone used sounds like space used by
processes other than ozone.
2. The meaning of capacity - remaining - used changes based on du.reserved. If
du.reserved is 0, capacity - remaining - used becomes space used by everything
other than the block files: this includes space used by both the datanode
metadata _and space used by other processes such as yarn._
All in all, my suggestion is to remove this term `non-ozone-used` from the UI
as its definition/meaning changes depending on configurations and so it's
confusing. We should go ahead with
https://issues.apache.org/jira/browse/HDDS-14483, which makes the current terms
clear: capacity -> ozone capacity, remaining -> ozone available etc. And also
introduces raw filesystem capacity, available, used etc.
> Non ozone used values are shown incorrectly with config changes
> ---------------------------------------------------------------
>
> Key: HDDS-14388
> URL: https://issues.apache.org/jira/browse/HDDS-14388
> Project: Apache Ozone
> Issue Type: Bug
> Components: Ozone Datanode, Ozone Recon
> Affects Versions: 2.0.1
> Reporter: Soumitra Sulav
> Assignee: Siddhant Sangwan
> Priority: Major
>
> On a fresh cluster with the default hdds.datanode.dir.du.reserved
> configuration, and we continue to write to non-Ozone directories, we will
> start seeing Non-Ozone used increasing as expected.
> On setting hdds.datanode.dir.du.reserved to a higher value, non ozone used is
> shown as 0 on cluster restart post the new configurations.
> This is also wrongly depicted in Recon, even when we have non-ozone used
> exactly the same.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
