rsync: dumping core, breaking pipes, not understanding modules - why?

Rob Furr rsfurr at
Fri Mar 1 06:08:48 EST 2002

I'm running rsync 2.5.2_1 on two machines - a production server (athlon 
1500+, 512mb, 160gb raid array, running FreeBSD 4.4-release) and a machine 
I'd like to use as a backup system (a p2/350, 192mb, 90gb, running FreeBSD 
4.5-release.)  I've got rsync running as a daemon on the production server, 
and I'd like to use rsync to back up the key directories on that machine to 
the backup system. However, it seems clear that I'm an idiot, and I'd 
massively appreciate any illumination as to how big an idiot I'm being:

I'm using the following in my file to initialize rsync as a 
daemon on startup on the production server:


/usr/local/bin/rsync --daemon --config=/etc/rsyncd.conf


This gets rsync going just fine (I assume.)

/etc/rsyncd.conf on the production server contains these lines:


log file = /var/log/rsyncd.log
pid file = /var/run/
lock file = /var/run/rsync.lock
uid = rsfurr # my own account - this will change once I get this working.
gid = web # the group that the files I want to back up are owned by
hosts allow = ___.___.___.___ #  the ip of the backup server

    comment = the web directories
    read only = yes

Here's the first problem: On the production server, typing rsync 
localhost:: results in this:

upeg# rsync localhost::
upeg_web        the upeg web directories!

So I'm reasonably sure that rsyncd.conf is being read and understood. 
However, when I do this (on the backup server):

rsync -a rsfurr at /bak/upeg_web/

I get this:

@ERROR: access denied to upeg_web from unknown ()
rsync: connection unexpectedly closed (62 bytes read so far)
rsync error: error in rsync protocol data stream (code 12) at io.c(140)

and when I do *this*:

rsync -a rsfurr at /bak/upeg_web/

I get *this*:

link_stat upeg_web/. : No such file or directory
client: nothing to do: perhaps you need to specify some filenames or the 
--recursive option?
rsync error: partial transfer (code 23) at main.c(23)

So, clearly, I'm either not issuing the right command (which I could have 
sworn I am), or I'm misconfiguring the module badly (which I don't *think* 
I am.)

Additionally, when I do *this:

rsync -a rsfurr at /bak/upeg_web/

it works  - it transfers all the files that rsfurr is allowed to read to 
the backup server ... but when it's done, it does this:

write failed flushing stdout buffer.
write stdout: Broken pipe
rsync error: partial transfer (code 23) at main.c(883)

There's two problems with this: while it does seem to get all the files, I 
was under the impression that setting up a module (e.g., [upeg_web] ) would 
restrict rsync access to just that module. If this isn't the case, that's 
okay, but I am a bit puzzled. The other problem is that I'd really like to 
not get error messages cluttering up my inbox when I've got this going on a 
regular basis. Also - the partial transfer worries me. Does this mean that 
at some point, the program might only partially transfer the last file?

On a related issue, I couldn't get ssh authentication working, for some 
reason  (ssh -v revealed that no matter what key format I had ssh-keygen 
make, it still fell through all the way down to password authentication, 
which made me testy, and I'm under a bit of a time crunch here, so I didn't 
want to pursue *that* problem right now...) so I installed expect and 
wrote/swiped-from-a-web-page a short little expect script to handle the issue:


#!/usr/local/bin/expect -f
log_user 0
spawn /usr/local/bin/rsync -a 
rsfurr at /bak/upeg_web/
expect "password: "
send "my_password_that_i'm_not_sharing_here/n"
log_user 1


(and yes, I've got this in /root with 700 permissions, and the backup 
server in question has no services at all running. Well, it's got ssh 
running, but once it goes into production, that's going off, and then it'll 
be a closed box except for the occasional rsync run.)

When I put this script in /etc/crontab in this way:

1	1	*	* 	* 	root	/foo/expect_script

rsync not only gives me the broken pipe error, but it also dumps core and 
leaves big hornkin' core dump files around. Why? Can I do anything about 
it? I've checked the faq-o-matic and the list archives on this issue (well, 
I've checked them for *all* these issues, and either they're not addressed, 
or I am, as I said before, an idiot.)

The reason I'm using rsync instead of a web crawler of some kind to do this 
backup is simple - there are other files on this server that need backup 
besides the web files. I'm just only worrying about getting the site backed 
up before I start worrying about the database directories and etc and var 
and like that.

Any suggestions as to where I went wrong will be, as I said at the top of 
this mess, definitely appreciated.

   Rob Furr
   rsfurr at

More information about the rsync mailing list