samba-bugs at samba.org
2016-Aug-15  10:18 UTC
[Bug 12153] New: Possible malfunctions of rsync after encountering >4GbFat32Error
https://bugzilla.samba.org/show_bug.cgi?id=12153
            Bug ID: 12153
           Summary: Possible malfunctions of rsync after encountering
                    >4GbFat32Error
           Product: rsync
           Version: 3.1.1
          Hardware: x86
                OS: Linux
            Status: NEW
          Severity: normal
          Priority: P5
         Component: core
          Assignee: wayned at samba.org
          Reporter: daniel-inet at gmx.net
        QA Contact: rsync-qa at samba.org
I use rsync to transfer a large number of files locally from one disk to
another.
rsync -avv --progress /media/source/ /media/destination/
source is on an ext4 filesystem 
destination is on a fat32 filesystem
The destination disk was checked and found to be OK by the badblocks-command
The destination filesystem was freshly created by mkdosfs.
There are chinese characters in the filenames. Filename encoding is utf-8 on
both source (defined by locales) and destination (reported by mount command).
Rsync ends up with an error message saying that a file from source is bigger
than 4gb. So far, this is expected behaviour because of the file size
limitation on a FAT32 filesystem.
Possible malfunction 1: Some other files smaller than 4gb are missing in the
destination as well.
Possible malfunction 2: When repeating the rsync command (to resume the
transfer), files are transferred that do already exist in the dest directory.
The files match in creation time and in comparison by diff command. fsck does
not report any errors. But nevertheless, some files are transferred again. I
can see big files being transferred again using the --progress option.
-- 
You are receiving this mail because:
You are the QA Contact for the bug.
samba-bugs at samba.org
2016-Aug-15  11:43 UTC
[Bug 12153] Possible malfunctions of rsync after encountering >4GbFat32Error
https://bugzilla.samba.org/show_bug.cgi?id=12153 --- Comment #1 from Daniel <daniel-inet at gmx.net> --- Possible malfunction 2 might be somehow related to case sensitivity in filenames. -- You are receiving this mail because: You are the QA Contact for the bug.
samba-bugs at samba.org
2016-Aug-15  12:43 UTC
[Bug 12153] Possible malfunctions of rsync after encountering >4GbFat32Error
https://bugzilla.samba.org/show_bug.cgi?id=12153 --- Comment #2 from Daniel <daniel-inet at gmx.net> --- Possible malfunction 2: Files are missing in destination. These files have been shown to be transferred with the --progress option at a speed that exceeds the maximum write speed of the disk on which the destination filesystem lays on. -- You are receiving this mail because: You are the QA Contact for the bug.
samba-bugs at samba.org
2016-Aug-15  14:01 UTC
[Bug 12153] Possible malfunctions of rsync when destination is on fat32 filesystem
https://bugzilla.samba.org/show_bug.cgi?id=12153 --- Comment #3 from Daniel <daniel-inet at gmx.net> --- Have been picking on the problem. I think that the problem with files which already exist in destination but are transferred again is not caused by rsync but by a time comparison mismatch which can occur when timestamp accuracy differs between source and destination filesystem. It would be nice to add something to the manpage, section "--modify-window", that this option makes sense when timestamp accuracy of source and destination filesystem differs. The reported problem remains that files are missing in the destination (and reported to have been transfered at a speed greater than possible) -- You are receiving this mail because: You are the QA Contact for the bug.
samba-bugs at samba.org
2016-Aug-15  14:40 UTC
[Bug 12153] Possible malfunctions of rsync when destination is on fat32 filesystem
https://bugzilla.samba.org/show_bug.cgi?id=12153 --- Comment #4 from Daniel <daniel-inet at gmx.net> --- This report might be a duplicate of 10675 -- You are receiving this mail because: You are the QA Contact for the bug.