I see the same problem on a linux box running without any network
involved. rsync -av dir1 dir2. (where dir[12] are both on local ide
disks)
I posted a few weeks ago regarding my 2.4.6 hang -- (not running
--deamon)
http://lists.samba.org/pipermail/rsync/2001-January/003552.html.
I found two 'fixes'.
1. run without the -v
2. or cntl-c and rerun the command to cleanup the rest of the tree
I still do a lot of solaris/linux rsyncs but I'm still using 2.3.2 on
those.
eric
Scott Russell wrote:
>
> All -
>
> I understand the focus of the discussion below is Linux <-> Solaris time
> outs but is anyone else seeing the same problem under Linux <-> Linux when
> running 2.4.6 rsync --deamon?
>
> Currently I'm seeing it from the client end. Both of the servers I'm pulling
> from were updated to 2.4.6 and then I started seeing problems. Since I don't
> have visibility to the server side there isn't to much else I can say for
> sure.
>
> On my end (client side) the system is Red Hat 6.0 + errata and a 2.2.18
> kernel using rsync 2.4.6. In my logs I see "unexpected EOF in read_timeout"
> usually right after the file list is received. Running rysnc with -v -v -v
> doesn't show anything special about the problem.
>
> -- Scott
>
> On Tue, Feb 27, 2001 at 01:29:45PM -0600, Dave Dykstra wrote:
> > Re: http://lists.samba.org/pipermail/rsync/2001-February/003628.html
> >
> > Some more email was exhanged the last couple days on the subject of a TCP
> > hang between Solaris and Linux:
> >
> > ------------------------------------------------------------------------
> >
> > > Date: Mon, 26 Feb 2001 12:06:08 -0600
> > > From: Dave Dykstra <[EMAIL PROTECTED]>
> > > To: "David S. Miller" <[EMAIL PROTECTED]>
> > > Cc: [EMAIL PROTECTED], [EMAIL PROTECTED]
> > > Subject: Re: Linux 2.2.16 through 2.2.18preX TCP hang bug triggered by rsync
> > >
> > > David,
> > >
> > > Regarding the patch you sent us (below) to try to help solve the problem of
> > > Solaris sending acks with a sequence number that was out-of-window:
> > >
> > > We have now completed extensive testing with it and analysis of tcpdumps
> > > and have determined that the patch is working as expected (accepting the
> > > acks) but it isn't enough to work around the state that Solaris gets itself
> > > into; the connection still hangs. It looks like Alexey was right. Linux
> > > is able to make further progress getting data sent to Solaris but it isn't
> > > enough to recover the whole session; the Linux receive window stays at 0 so
> > > I presume the rsync application isn't reading the data because it's waiting
> > > for the Solaris side to complete something. Oddly, every 30 or 60 seconds
> > > after this situation occurs, Linux sends another 2-4 1460 byte packets and
> > > they're acknowledged by Solaris. It seems unlikely that the rsync
> > > application would be sending exact multiples of 1460, but I didn't do a
> > > trace during the hang to see if it was generating extra data for some
> > > reason.
> > >
> > > I have attached the tcpdump in case you're interested. Recall that 'static'
> > > is Linux and 'dynamic' is Solaris. We have added our interpretation on
> > > some of the lines.
> > >
> > > We also have had an initial response from Sun where they recommended
> > > upgrading with a certain patch but that too hasn't solved the problem (the
> > > attached tcpdump is with the Solaris patch in place).
> > >
> > > Thanks for your help and I'll let you know if we do ever get a satisfactory
> > > answer from Sun.
> > >
> > > - Dave Dykstra
>
> --
> Regards,
> Scott Russell ([EMAIL PROTECTED])
> Linux Technology Center, System Admin, RHCE.
> T/L 441-9289 / External 919-543-9289
> http://bzimage.raleigh.ibm.com/webcam
--
__________________________________________________________________
Eric T. Whiting AMI Semiconductors
(208) 234-6717 2300 Buckskin Road
(208) 234-6659 (fax) Pocatello,ID 83201
[EMAIL PROTECTED]
