Rsync fails on local transfers

Cameron Simpson cs at zip.com.au
Wed Jul 18 02:03:25 UTC 2007


On 17Jul2007 11:16, Justin W <jlist at jdjlab.com> wrote:
> Justin W wrote:
>> I've got a script I wrote up which backs up my computer using rsync.  A
>> short overview is that I created a file system image using dd,

Don't you create filesystems with mkfs/e2mkfs ?
You need to use dd to make a file to _hold_ the filesystem, but that's
just a chunk of data; it's not a filesystem until you use mkfs.

>> and now I
>> mount the image and rsync from the server's HD to the image file.
>>
>> Periodically during large backups, I get the following error:
>>
>>     rsync: connection unexpectedly closed (16931 bytes received so far)
>>     [sender]
>>     rsync error: error in rsync protocol data stream (code 12) at
>>     io.c(453) [sender=2.6.9]

No other messages?

>> When I Google for the problem, I get many results about a SSH connection
>> timing out because of inactivity during the delete stage, but I'm not
>> using SSH, so I don't get why I would have this problem: both the source
>> and destination are local.

Correct.

>> Any ideas on how to debug this would be great. I've thought of using
>> strace, but it'd result in a huge file which would have to be sorted
>> through (as rsync isn't technically crashing, rsync would continue with
>> a shutdown procedure and the logging wouldn't stop right at the point of
>> failure).

Half of rsync is crashing; it forks. Run "strace -f", and then look at the
pid that _isn't_ the one printing the error message.

> Actually, looking through my script, I had stuck an strace in there at some 
> point.  I currently have a 165MB trace in my /tmp directory.  Here's the 
> last few lines of it:
[...]
>    write(1, "usr/lib/locale/nb_NO/LC_IDENTIFI"..., 39) = 39
[...]
>    select(6, [5], [], NULL, {60, 0})       = 1 (in [5], left {60, 0})
>    read(5, "", 4)                          = 0
>    write(2, "rsync: connection unexpectedly c"..., 76) = 76
>    write(2, "\n", 1)                       = 1
[...]
>> Attached is the backup script I use (the configuration file it loads in
>> is very simple, so I'm not including that).

Regrettably, that's just the code that prints the error message. See if
"strace -f" is more informative.
-- 
Cameron Simpson <cs at zip.com.au> DoD#743
http://www.cskk.ezoshosting.com/cs/

These aren't the droids you're looking for.




More information about the fedora-list mailing list