Protocol stream error on extended attribute, silent failure to copy all attributes

[Matt Rosenberg]

> From: Henri Shustak <henri.shustak at gmail.com>
> Date: Thu, 3 Feb 2011 00:48:02 +1300
> To: rsync <rsync at lists.samba.org>
> Subject: Re: Protocol stream error on extended attribute, silent failure to
> copy all attributes
>
> [snip]
> 
> Recently, a user posted the following rsync error to the LBackup mailing list
> : 
> 
>> "rsync: writefd_unbuffered failed to write 4 bytes to socket [sender]: Broken
>> pipe (32)
>> rsync: connection unexpectedly closed (3225054 bytes received so far)
>> [sender]
>> rsync error: error in rsync protocol data stream (code 12) at io.c(601)
>> [sender=3.0.7]"
>> 
>> Can anyone tell me what it means, why it would occur and most importantly how
>> to fix it.
>
> In their case the destination volume disk was running out of space. However,
> for some reason rsync was not reporting this error as I would expect.
> 
> A link to this thread on the LBackup mailing is available below :
>  http://tinyurl.com/lbackup-discussion-diskfull
> 
> I am in the process of attempting to reproduce this error when the disk is
> full. During testing I have cared out to date, I have not been able to
> reproduce rsync results error output when the disk is full. In stead I receive
> something like that which is quoted below.

Hi Henri,

In the setup I used to produce my example, I have almost 200 GB of available
space on the destination volume (the source has almost 600). I was only
copying a few hundred MB.

On my production servers where I originally encountered this problem, the
free space on each volume is over 1 TB.

Most of the time, I get the protocol stream error. The writed_unbuffered one
only occurs once out of maybe 5 or 6 runs.

Thanks for your reply,

Matt