ArafatKhan2198 opened a new pull request, #8836:
URL: https://github.com/apache/ozone/pull/8836
## What changes were proposed in this pull request?
When a directory and its files are deleted, fetchSizeForDeletedDirectory
returns incorrect size calculations because it relies solely on
NamespaceSummary data, which doesn't account for files that still exist in
deletedTable.
Steps to Reproduce:
```
1. Create directory structure: /dir1/dir2/f1.txt
2. Delete dir2 (Event 1: directory deletion)
3. Delete f1.txt (Event 2: file deletion)
4. Call fetchSizeForDeletedDirectory for dir2
```
Expected: Returns actual disk space consumed by f1.txt in deletedTable
Actual: Returns 0 because NamespaceSummary.sizeOfFiles was subtracted
Root Cause: NamespaceSummary tracks deleted files by subtracting their size,
but deletedTable still contains the actual data blocks.
fetchSizeForDeletedDirectory only considers NamespaceSummary data, leading to
incorrect size calculations.
Impact: Incorrect space reclamation reports and misleading storage analytics.
This pull request updates the Recon NamespaceSummary logic to correctly
account for deleted files and directories, ensuring accurate size reporting for
deleted directories.
### **Key code changes include:**
- **NamespaceSummary Model & Codec:**
- Extended the `NSSummary` class and its codec to track the number and
size of deleted files and directories, as well as the set of deleted child
directories.
- Ensured that serialization and deserialization of `NSSummary` objects
now include these new deleted-tracking fields, with backward compatibility for
older data.
- **Event Handling Logic:**
- Updated the event processing in `NSSummaryTaskWithFSO` to properly
update the deleted file and directory tracking fields in `NSSummary` when
deletion events are received.
- When a file is deleted, its size and count are now added to the
deleted tracking fields of the parent directory’s `NSSummary`, rather than
simply subtracting from the active file size/count.
- When a directory is deleted, its object ID is added to the parent’s
`deletedChildDir` set, and the deleted directory count is incremented.
- **Size Calculation Consistency:**
- Ensured that methods which fetch the size for deleted directories
(such as `fetchSizeForDeletedDirectory`) now consider both the deleted files
and directories, preventing under-reporting of disk usage when files are
deleted after their parent directory.
- **Backward Compatibility:**
- The codec and event handling logic are designed to be backward
compatible, so existing data and event flows will not break.
## What is the link to the Apache JIRA
https://issues.apache.org/jira/browse/HDDS-13479
## How was this patch tested?
Tested out manually
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
