rsync - Sep 2008 - DO NOT REPLY [Bug 5795] New: error in rsync protocol data stream (code 12) at io.c(632) [sender=3.0.4]

https://bugzilla.samba.org/show_bug.cgi?id=5795

           Summary: error in rsync protocol data stream (code 12) at
                    io.c(632) [sender=3.0.4]
           Product: rsync
           Version: 3.0.4
          Platform: x86
        OS/Version: Mac OS X
            Status: NEW
          Severity: blocker
          Priority: P3
         Component: core
        AssignedTo: wayned@samba.org
        ReportedBy: Dave@Yost.com
         QAContact: rsync-qa@samba.org


0 424 Z% rsync -av --sparse --progress --partial 
/Users/yost/Documents/VMWare/winxp.vmwarevm  /Volumes/x/vmware
sending incremental file list
winxp.vmwarevm/
winxp.vmwarevm/winxp.vmdk
  1578172416  17%   11.08MB/s    0:10:35
rsync: writefd_unbuffered failed to write 4 bytes [sender]: Broken pipe (32)
rsync: connection unexpectedly closed (32 bytes received so far) [sender]
rsync error: error in rsync protocol data stream (code 12) at io.c(632)
[sender=3.0.4]
12 425 Z% rsync --version
rsync  version 3.0.4  protocol version 30
Copyright (C) 1996-2008 by Andrew Tridgell, Wayne Davison, and others.
Web site: http://rsync.samba.org/
Capabilities:
    64-bit files, 32-bit inums, 32-bit timestamps, 64-bit long ints,
    socketpairs, hardlinks, symlinks, IPv6, batchfiles, inplace,
    append, ACLs, xattrs, iconv, symtimes

rsync comes with ABSOLUTELY NO WARRANTY.  This is free software, and you
are welcome to redistribute it under certain conditions.  See the GNU
General Public Licence for details.
0 426 Z%


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

https://bugzilla.samba.org/show_bug.cgi?id=5795





------- Comment #1 from Dave@Yost.com  2008-09-27 18:23 CST -------
It turns out there was clearly not enough space on the destination. The copy
was going to take 20GB, and after the crash there was still 1GB on the
destination, so I thought this was probably a protocol bug rather than a
problem with running out of space.

Perhaps what this is is a situation where the remote end should have sent back
an indication that it was out of space and should have shut down gracefully.

But there is another problem here: the two ends should start out by negotating
whether there is enough space for the copy. In a better world, the destination
OS would give the rsync process the ability to atomically grab disk resources
up front to be used for the files and folders it creates, and if that fails,
the remote rsync would tell the UI rsync no dice. Or how about this: writing
the destination files could be transaction in the OS file system! Nah.


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

https://bugzilla.samba.org/show_bug.cgi?id=5795


wayned@samba.org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Severity|blocker                     |enhancement
             Status|NEW                         |ASSIGNED
            Summary|error in rsync protocol data|look into a improved tear-
                   |stream (code 12) at         |down processing during fatal
                   |io.c(632) [sender=3.0.4]    |errors




------- Comment #2 from wayned@samba.org  2008-10-02 13:53 CST -------
Yes, the error reporting coming back from some errors can indeed be lacking. 
However, the pipe-lined nature of the protocol can make this hard to overcome
(the error can be behind so much checksum data that it can't make it back
prior
to the connection getting torn down).  In 3.1.0, I have a new option,
--msgs2stderr, that can often be used to debug such situations (for non-daemon
transfers).

It would be good to investigate a reliable way to drain (and discard) the
pending data to get all the relevant messages more reliably. For instance, if a
new message was added "fatal exit in progress", it could be sent and
circle the
3 processes before the connection is torn down.  e.g. a write error on the
receiver sends the error message (text) to the generator, sends the fatal
message too, and then just discards file-change data until it gets the fatal
message back from the sender.


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

https://bugzilla.samba.org/show_bug.cgi?id=5795





------- Comment #3 from Dave@Yost.com  2008-10-02 15:47 CST -------
Why not have the process generating the error delay exit until it gets an ack
of the error message? A separate connection would be best so the pipelined data
doesn't get in the way.


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

https://bugzilla.samba.org/show_bug.cgi?id=5795





------- Comment #4 from wayned@samba.org  2008-10-03 00:35 CST -------
That's what I just said.


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

https://bugzilla.samba.org/show_bug.cgi?id=5795





------- Comment #5 from matt at mattmccutchen.net  2009-11-27 20:45 CST -------
This is fixed in the current development rsync, right?  When it gets an error
writing to the socket, it continues to read for messages from the other side.


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

https://bugzilla.samba.org/show_bug.cgi?id=5795

Wayne Davison <wayned at samba.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |RESOLVED
            Version|3.0.4                       |3.1.0
         Resolution|                            |FIXED

--- Comment #6 from Wayne Davison <wayned at samba.org> 2011-12-24
22:18:04 UTC ---
The upcoming 3.1.0 tries to ensure that it drains the errors from the transfer,
so hopefully this is fixed for you.

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