Shilun Fan created HDDS-15628:
---------------------------------
Summary: [DiskBalancer] Preserve report output order for datanodes
with the same density
Key: HDDS-15628
URL: https://issues.apache.org/jira/browse/HDDS-15628
Project: Apache Ozone
Issue Type: Sub-task
Reporter: Shilun Fan
Assignee: Shilun Fan
DiskBalancer report command sorts datanodes by currentVolumeDensitySum in
descending order. However, when multiple datanodes have the same
currentVolumeDensitySum, the output order may not be stable.
The current implementation builds the report list from a ConcurrentHashMap
values collection before sorting. ConcurrentHashMap does not preserve insertion
order, so datanodes with equal density may be displayed in an order different
from the user-provided datanode order.
For example, if the user runs the report command with:
ozone admin datanode diskbalancer report host-2 host-1
and both host-2 and host-1 have the same currentVolumeDensitySum, the output
may show host-1 before host-2.
The report output should remain deterministic and preserve the user input order
when density values are equal. This makes the CLI behavior easier to understand
and avoids unnecessary output instability.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
