problem from NeXTStep 3.3 rlogin to SunOS 4.1.4

problem from NeXTStep 3.3 rlogin to SunOS 4.1.4

[Timothy J. Luoma]

A recently developed problem which I cannot explain.

When I rlogin to my SUN account from my NeXT, I have been "losing"  
output.  Most often it happens with 'ls' or 'cat' where the command  
runs but nothing comes to the screen.  If I feed it to 'less' that  
works.

The NeXT is running NeXTStep 3.3 pl1 and I'm using Stuart.app.

The rlogin connection is over PPP connection (mtu/mru 1500, 28.8  
speed, serial port at 57600)

The Sun machine is running 4.1.4 and zsh 3.0.0

Any clues?

TjL

--
Tj Luoma (luomat@peak.org)	PEAK OpenStep/NeXTStep FTP Site
Newly revised and enhanced NeXTStep/OpenStep resources page, over
300 NeXTStep/OpenStep/Rhapsody related links.
	http://www.peak.org/~luomat/next/

Re: problem from NeXTStep 3.3 rlogin to SunOS 4.1.4

[Bart Schaefer]

On Mar 4,  2:04pm, Timothy J. Luoma wrote:
} Subject: problem from NeXTStep 3.3 rlogin to SunOS 4.1.4
}
} When I rlogin to my SUN account from my NeXT, I have been "losing"  
} output.  Most often it happens with 'ls' or 'cat' where the command  
} runs but nothing comes to the screen.

Do you really mean "rlogin"?  As in, you log in to the Sun, get a prompt,
and then start issuing "ls" and "ls | less" commands?

Or do you mean "rsh" as in "rsh sun ls" vs. "rsh sun ls | less"?

} If I feed it to 'less' that works.
} 
} Any clues?

If it's "rsh" then this is (or used to be) a common problem.  It happens
when the rsh process on the local side exits before the output buffered
in the network layer gets flushed to the screen.

The local rsh exits when it gets notified on the exit-status connection
that the remote process has terminated; that may leave unread data on
the stdout/stderr connections.  If some other process (like less) is
waiting for the output, or if the output is sufficiently large, the
buffers get flushed before local rsh exits (before the shell gets control
back again).

I don't know why it would have begun happening to you now if it wasn't
before, but in my experience the problem isn't limited to zsh.

-- 
Bart Schaefer                             Brass Lantern Enterprises
http://www.well.com/user/barts            http://www.nbn.com/people/lantern

Re: problem from NeXTStep 3.3 rlogin to SunOS 4.1.4

[Andrej Borsenkow]

On Tue, 4 Mar 1997, Timothy J. Luoma wrote:

> 
> When I rlogin to my SUN account from my NeXT, I have been "losing"  
> output.  Most often it happens with 'ls' or 'cat' where the command  
> runs but nothing comes to the screen.  If I feed it to 'less' that  
> works.
> 

I can only confirm, that it happens. It happens with _rlogin_ (and not
with _rsh_). It is not limited to NeXT. I have two SVR4 based systems -
one is based on original SVR4.0.3 for i386 (SINIX-L), other on DC/OSx from
Pyramid (SINIX-N).

rlogin SINIX-N -> SINIX-L works O.K. 
rlogin SINIX-L -> SINIX-N shows the above problem. It is not always the
whole output lost - sometimes I see only the first line of `ls' output;
sometimes the `CR NL' after prompt is lost, that is:
% ls
 results in
% ls....... (first line of ls output).

It seems to depend on output size - `ls -l' have no problems.

`cat' OTOH works. 

I actually have no clues, except supposing some bugs in system. If anybody
have the ideas, how this could be debugged - much thanks in advance.

greetings

-------------------------------------------------------------------------
Andrej Borsenkow 		Fax:   +7 (095) 252 01 05
SNI ITS Moscow			Tel:   +7 (095) 252 13 88

NERV:  borsenkow.msk		E-Mail: borsenkow.msk@sni.de
-------------------------------------------------------------------------

Re: problem from NeXTStep 3.3 rlogin to SunOS 4.1.4

[Zoltan T. Hidvegi]

Andrej Borsenkow wrote:
> > When I rlogin to my SUN account from my NeXT, I have been "losing"
> > output.  Most often it happens with 'ls' or 'cat' where the command
> > runs but nothing comes to the screen.  If I feed it to 'less' that
> > works.
> >
>
> I can only confirm, that it happens. It happens with _rlogin_ (and not
> with _rsh_). It is not limited to NeXT. I have two SVR4 based systems -
> one is based on original SVR4.0.3 for i386 (SINIX-L), other on DC/OSx from
> Pyramid (SINIX-N).
>
> rlogin SINIX-N -> SINIX-L works O.K.
> rlogin SINIX-L -> SINIX-N shows the above problem. It is not always the
> whole output lost - sometimes I see only the first line of `ls' output;
> sometimes the `CR NL' after prompt is lost, that is:
> % ls
>  results in
> % ls....... (first line of ls output).

Do you use setopt no_flow_control?  If yes, did you try to unset it?

Zoltan

Re: problem from NeXTStep 3.3 rlogin to SunOS 4.1.4

[Andrej Borsenkow]

On Wed, 5 Mar 1997 hzoli@VNET.IBM.COM wrote:

> > rlogin SINIX-L -> SINIX-N shows the above problem. It is not always the
> > whole output lost - sometimes I see only the first line of `ls' output;
> > sometimes the `CR NL' after prompt is lost, that is:
> > % ls
> >  results in
> > % ls....... (first line of ls output).
> 
> Do you use setopt no_flow_control?  If yes, did you try to unset it?
> 

thousand thanks! running with
  `unsetopt no_flow_control'
really helps (I havn't set it - it must be on by default). 

It is probably still OS bug (happens with rlogin only - telnet runs O.K.),
but as long as workaround exists I don't care.

greetings

-------------------------------------------------------------------------
Andrej Borsenkow 		Fax:   +7 (095) 252 01 05
SNI ITS Moscow			Tel:   +7 (095) 252 13 88

NERV:  borsenkow.msk		E-Mail: borsenkow.msk@sni.de
-------------------------------------------------------------------------

Re: problem from NeXTStep 3.3 rlogin to SunOS 4.1.4

[Bart Schaefer]

On Mar 5,  9:47am, Andrej Borsenkow wrote:
} Subject: Re: problem from NeXTStep 3.3 rlogin to SunOS 4.1.4
}
} rlogin SINIX-L -> SINIX-N shows the above problem. It is not always the
} whole output lost - sometimes I see only the first line of `ls' output;
} sometimes the `CR NL' after prompt is lost

On Mar 5, 10:54am,  (Zoltan T. Hidvegi) wrote:
} Subject: Re: problem from NeXTStep 3.3 rlogin to SunOS 4.1.4
}
} Do you use setopt no_flow_control?  If yes, did you try to unset it?

On Mar 5,  7:09pm, Andrej Borsenkow wrote:
} Subject: Re: problem from NeXTStep 3.3 rlogin to SunOS 4.1.4
}
} thousand thanks! running with
}   `unsetopt no_flow_control'
} really helps (I havn't set it - it must be on by default). 

Did you unset this in the local shell before starting rlogin, or in the
remote shell afterwards?  (I suspect the remote one.)

} It is probably still OS bug (happens with rlogin only - telnet runs O.K.),
} but as long as workaround exists I don't care.

This begins to sound more and more familiar.  Is the stty line speed
slower on the local terminal where you started rlogin than it is on the
remote psuedo-tty?  If so, you might try forcing them to be the same to
see if that also makes the problem go away.

The rlogin processes probably want to use flow control to compensate for
the different line speeds.

-- 
Bart Schaefer                             Brass Lantern Enterprises
http://www.well.com/user/barts            http://www.nbn.com/people/lantern

Re: problem from NeXTStep 3.3 rlogin to SunOS 4.1.4

[Timothy Luoma]

> 
> Do you use setopt no_flow_control?  If yes, did you try to unset it?
 
It is on.  I have not had the chance to test it and see if turning it off
would help, but I suspect it will given that someone else reports it
working for them under similar situations.

TjL