[Bug 10290] New: Regression since 3.0.9: send_xattr_request: Assertion `f_out >= 0' failed

[samba-bugs at samba.org]

https://bugzilla.samba.org/show_bug.cgi?id=10290

           Summary: Regression since 3.0.9: send_xattr_request: Assertion
                    `f_out >= 0' failed
           Product: rsync
           Version: 3.1.0
          Platform: x64
        OS/Version: Linux
            Status: NEW
          Severity: regression
          Priority: P5
         Component: core
        AssignedTo: wayned at samba.org
        ReportedBy: moritz at bunkus.org
         QAContact: rsync-qa at samba.org


Hey,

I'm running daily backups with dirvish (which uses rsync with --link-dest, very
similar to rsnapshot). This creates one directory per day for the machine I'm
backing up. It works fine with both rsync 3.0.9 and 3.1.0.

Once a week I also sync the whole machine directory containing those daily
directories to an extern hard drive connected via USB. This weekly sync has
been working fine for years with rsync 3.0.9. With 3.1.0 I consistently receive
the error message mentioned in $subject.

Here's a copy&paste from the rsync invocation including the error message:

[0 root at sweet-chili ~] date ; time ionice -c3 ~mosu/opt/rsync/bin/rsync -haxHAX
--delete /backup/ /backup/ext/full1/backup/ ; date
Sa 23. Nov 14:52:35 CET 2013
rsync: xattrs.c:555: send_xattr_request: Assertion `f_out >= 0' failed.
rsync: [receiver] write error: Broken pipe (32)
rsync error: error in socket IO (code 10) at io.c(823) [receiver=3.1.0dev]
rsync: [sender] write error: Broken pipe (32)
rsync error: error in socket IO (code 10) at io.c(823) [sender=3.1.0dev]
ionice -c3 ~mosu/opt/rsync/bin/rsync -haxHAX --delete /backup/   1116,40s user
4586,98s system 18% cpu 8:37:24,14 total
Sa 23. Nov 23:29:59 CET 2013

I've also spent the last couple of weeks bisecting the exact commit which
caused this breakage. That's why the above paste contains that version number.
The first git revision to trigger the issue is
78286a03d196efeed754aa4c1faa571f5dbea342.

All following revisions that I've tried (even some from git after the 3.1.0
release) are consistent in that they abort with the aforementioned error.

Some more facts about my system:

* Arch Linux 64bit, updated pretty much weekly
* Source file system is btrfs mounted with »noatime,nodiratime« options
* Destination file system is btrfs mounted with
»compress=lzo,noatime,nodiratime« options

I'm willing to provide any further information you may require including
re-runs with different options, but please keep in mind that each run takes
anywhere between 7 and 12 hours depending on when the previous run was done,
how much data has changed in the meantime, and whether or not it aborts.

-- 
Configure bugmail: https://bugzilla.samba.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug.