Daojiang Yan created KUDU-3797:
----------------------------------
Summary: kudu table statistics: use human-readable byte format
consistently for on-disk size
Key: KUDU-3797
URL: https://issues.apache.org/jira/browse/KUDU-3797
Project: Kudu
Issue Type: Improvement
Reporter: Daojiang Yan
Assignee: Daojiang Yan
In
[tool_action_table.cc|vscode-file://vscode-app/Applications/Visual%20Studio%20Code.app/Contents/Resources/app/out/vs/code/electron-browser/workbench/workbench.html],
the {{GetTableStatistics}} action prints the table-level {{on_disk_size}} as a
raw byte count via {{{}std::to_string(){}}}, while the per-tablet
{{on_disk_size}} (printed when {{--show_tablets}} is set) uses
{{{}HumanReadableNumBytes::ToString(){}}}. As a result, the same value can
appear as {{on disk size: 3145728}} at the table level and {{3.0M}} in the
tablet rows.
The raw format was intentionally kept for backward compatibility with the
existing {{KuduTableStatistics::ToString()}} output. This ticket tracks making
the byte formatting consistent across both places (e.g. by using
{{HumanReadableNumBytes::ToString()}} for the table-level fields as well, or
otherwise reconciling the two).
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
