On Fri, 5 Nov 2004 00:29:21 -0600, Bob Tanner <tanner at real-time.com> wrote:
> For months I have been getting the following error when attempting to mirror
> fedora core stuff:
> 
> rsync: writefd_unbuffered failed to write 4092 bytes: phase
> "unknown" [generator]: Connection reset by peer (104)
> rsync error: received SIGUSR1 or SIGINT (code 20) at main.c(965)
> rsync error: error in rsync protocol data stream (code 12) at io.c(909)
> 
> Before FC, I was getting the same error for consumer redhat (7.3, 8.0, 9,
> etc).
> 
> ftpadmin at redhat.com is about as helpful as a headache and treated me like a
> clueless smuck. Public comment! ftpadmin at redhat.com = big ass dork! :-P
> 
> First solution was upgrade rsync.
> 
> $ rsync --version
> rsync  version 2.6.3  protocol version 28
> 
> I always run the latest stable version.
> 
> Next was ACL(?) problem my side, but they could not explain how I can get some
> files, before the error.
> 
> Next was QoS in our core router. Uhh, what QoS?  Don't do any QoS for this
> segment in the core router.
> 
> Next was vlan issue, it's not on a vlan issue.
> 
> Next was load balancer problem their side (ah!) first intelligent response.
> 
> Never head anything about it.
> 
> Followed up, a couple weeks later, got rsync version speech. Groan.
> 
> Sent the email from ftpadmin about LB problem. Silence.
> 
> Followed up again. Silence.
> 
> More or less gave up and did this:
> 
> while (true); do  ~/bin/rsync-fedora-redhat ; sleep 5m; done
> 
> Eventually(!) I get everything but only days, if not weeks after the release
> onto the download master mirrors.
> 
> Most of the helpful troubleshooting advise
> (http://rsync.samba.org/issues.html) is for the server side of the
> connection. So, I'm at a loss what to do next.
> 
> Help?
> 

Just for the hell of it, have you tried using rsync from source or
debian or some other place?

Something we do for mirroring debian is split it into a few jobs, that
might help some if they have a bad LB in the middle.

I also remember seeing this issue before when I had a bad (=cheap)
network card on the server end. It had trouble with most anything that
pushed high bandwith.  I assume you would have a good nic on your end,
but if you run out of things to do while idle() you could double check
it I guess.

Jay

-- 
Jay Kline
http://www.slushpupie.com/

_______________________________________________
TCLUG Mailing List - Minneapolis/St. Paul, Minnesota
Help beta test TCLUG's potential new home: http://plone.mn-linux.org
Got pictures for TCLUG? Beta test http://plone.mn-linux.org/gallery
tclug-list at mn-linux.org
https://mailman.real-time.com/mailman/listinfo/tclug-list