Bug 15567 - Data inconsistency with vdbench
Summary: Data inconsistency with vdbench
Status: NEW
Alias: None
Product: Samba 4.1 and newer
Classification: Unclassified
Component: VFS Modules (show other bugs)
Version: 4.19.3
Hardware: x86 Linux
: P5 normal (vote)
Target Milestone: ---
Assignee: Samba QA Contact
QA Contact: Samba QA Contact
Depends on:
Reported: 2024-02-02 01:38 UTC by sharingfun520@outlook.com
Modified: 2024-02-02 01:38 UTC (History)
0 users

See Also:


Note You need to log in before you can comment on or make changes to this bug.
Description sharingfun520@outlook.com 2024-02-02 01:38:24 UTC
I use vdbench test samba for 12h.There’s no error in log.smbd.samba 4.19.3 is configured as “security=ADS” and the other were default (by default,aio read size =1).
These two directories are under the same filesystem and are shared by the same user.Although vdbench reported an error, the final file written to disk was correct. The behavior of vdbench is to read the contents immediately before writing to verify that the last write was correct, which may be from the cache. samba aio whether there is a mechanism such as caching, resulting in data is not strongly consistent. 
The vdbench reports that

“Data Validation error for fsd=fsd2;FSD lba: 0x04ce2409;Key block size:8192; relative sector in data block: 0x02
File nme:/test/a2/wdb.1_2.dir/vdb.2_2.dir/wdb_f0076.file; file block lba: 0x000e2000; bad sector file lba: Cx000e2400
==-> Data validation Key miscompare.Expecting key=0x3e received key=x3a (62/58)
=>Logical byte address miscompare. Expecting 0x04ce2400,receiving 0x18de2400
-e> Bad owner id. Expecting 0x4e7a (20090) but found 0x10ce
There are no mismatches in bytes 32-511”

The vdbench testfile is
” data_error=1

Meanwhile,the smbclient is linux that kernel is "3.10.0-514.el7.x86_64",just only test "/test11/cifs2" reported such error.