Re: Protocol error: uncounted data discarded

[John Minnihan]

Hi Chris,

I have seen this too, but not consistently.  For me, the error occurs only
near the end of a very large checkout using a 'universe' defining module
name.  The error does not exhibit during smaller module checkouts, and as I
said - it does not consistently show up even during the large checkouts.

I'm curious what you see as the difference between 'in a buld script' and
'by hand'.

My WAG at this point is that the error shows up only when my network is
under heavy load.  During periods of relative network 'quiet' (say, 15 -
40% bandwidth utilization), the large checkouts complete as expected.

HTH you in troubleshooting and developing a workaround.  I posted my
question (with details) about a month ago.  No reply yet, and the only
other mention to this in the archives was your message of 4-MAY-2001.  See
my message of 20-JUN-2001 RE: that.

address@hidden wrote:

> Has there been any update on this?  I made the mistake of upgrading
> one of my servers and one of my build machines to 1.11.1p1
> and now I get this whenever I try a build using a build script:
>
>  /usr/bin/cvs -q co -r build-1266 tree3
> Protocol error: uncounted data discarded
>
> The client is a Linux RH 6.0 2.2.17 box.
> The server is a Linux RH 6.2 2.2.14-5.0smp box.
>
> Doing the same thing by hand seems to always work.
>
> Thanks...
>
> chuckr
>
> On Fri, May 04, 2001 at 04:23:14PM -0400, Chris Sharpe wrote:
> > I just upgraded by Solaris (sparc) pserver to cvs 1.11.1p1 and from
> > a WinNT command line cvs 1.11 client, I sometimes get the following
> > error:
> >
> >    cvs -d $my_cvsroot -f -r -q -z3 checkout -r $my_tag $my_module
> >    Protocol error: uncounted data discarded
> >
> > The module is about 3000 files in size and I've seen this a couple
> > of times when running the checkout command as part of a script.
> > Other modules checked out immediately after the failure work fine
> > and if I try to checkout this module again, it works fine.
> >
> > I saw that there was a FIXME comment recently added by Derek in
> > server.c that might be related.  Anyone have any insight?  Is this
> > error related to 1.11.1p1 or is the timing just a coincidence
> > (I never saw this error with 1.11 pserver).
> >
> > +-------------------------------------------------------------------+
> > | Chris Sharpe  KF4WVO         Author of "The Unofficial PEZ FAQ"   |
> > | address@hidden   "PEZ - A treat to eat in a toy that's neat"|
> > +-------------------------------------------------------------------+
>
> _______________________________________________
> Info-cvs mailing list
> address@hidden
> http://mail.gnu.org/mailman/listinfo/info-cvs

--
_____________________________________
John Minnihan
mailto:address@hidden
http://www.freepository.com