xon/xoff problem

xon/xoff problem

Post by Joel Willia » Wed, 20 Mar 1996 04:00:00



Unix 3.2v4.2 to Fujitsu line printer xon/xoff interface

We have a Fujitsu line printer running at 9600 buad with xon/xoff
interface into an Arnet multiport box.  Lately we've been experiencing
strang handshaking related issues (either printer gets buffer overflow
or print job stops in middle, printer still online, job still in
queue.)  I'm using the 'dumb' interface and spooling everything.  My
stty line in the interface script is:

stty 9600 -parenb cs8 ixon ixoff -ixany clocal 0<&1

I've eliminated the cable (four wire, flat), connector and Arnet port
by using a different port cable and connector with the same strange
errors.  The old addage 'boot and try again' seems to help as after
booting, the printer works fine for a week or two and then things go
south.

The original installers of the system setup stty lines in the
/etc/rc2.d/S80lp file for each 'printer' port.  Reading the Unix doc,
that appears only necessary for direct printing.  I'm always spooling.

Is there anything I'm missing?

Joel Williams
MIS Director
Physicians' MSO


 
 
 

xon/xoff problem

Post by sy.. » Thu, 21 Mar 1996 04:00:00


In the beginning, or at least on 19 Mar 96 00:43:49 GMT,  Joel Williams

Quote:

>strang handshaking related issues (either printer gets buffer overflow
>or print job stops in middle, printer still online, job still in
>queue.)  I'm using the 'dumb' interface and spooling everything.  My
>stty line in the interface script is:

>stty 9600 -parenb cs8 ixon ixoff -ixany clocal 0<&1

--

It's klunky, but I use the following for some of my
finicky printers

stty 9600 -parenb cs8 ixon -ixany clocal ff1 cr2 nl0 0<&1

This usually gets it.

 -------------------------------------------------


 Calcasieu Lumber Co.           uunet!calcas!sysop
 Austin, TX                                        
 --  Windows for the users;  UNIX for the work  --

 
 
 

xon/xoff problem

Post by Jeff Lieberma » Tue, 02 Apr 1996 04:00:00


: We have a Fujitsu line printer running at 9600 buad with xon/xoff
: interface into an Arnet multiport box.  Lately we've been experiencing
(...)

: stty 9600 -parenb cs8 ixon ixoff -ixany clocal 0<&1
That should work.  I would suggest:
        stty 9600 ixon ixoff -ixany -parenb cs8 clocal onlcr opost 0<&1
I believe that the opost is required or no post processing occurs.

: I've eliminated the cable (four wire, flat), connector and Arnet port
: by using a different port cable and connector with the same strange
: errors.

I don't believe it.  Your setup is fine, but it exhibits all
the classic symptoms of handshake failure.  Here's how you test it:
        Print a long document to the printer.  /etc/termcap is fine.
        Install a breakout box (with lights) on the serial port.
        Stomp on the ON-LINE button randomly.
        Data flow should stop when the printer is off-line and
                resume when it is returned on-line.
My gut feel is that the TD wire (that which sendeth the xon and xoff
characters back to Arnet) is broken.

: The old addage 'boot and try again' seems to help as after
: booting, the printer works fine for a week or two and then things go
: south.

I suspect your printer has a large buffer and will work just fine
for jobs that fit.  My guess is that it only shows up for large jobs.

: The original installers of the system setup stty lines in the
: /etc/rc2.d/S80lp file for each 'printer' port.  Reading the Unix doc,
: that appears only necessary for direct printing.  I'm always spooling.

This is un-necessary unless you're sending print jobs directly to
the port.  I suggest you get rid of these.

[x] Email to author  [ ] To mailing list  [x] Posted to newsgroup

--
# Jeff Liebermann  Liebermann Design  150 Felker St #D  Santa Cruz  CA  95060

# 408.699.0483 digital_pager    73557,2074  cis [don't]

 
 
 

xon/xoff problem

Post by Tony Lawren » Wed, 03 Apr 1996 04:00:00


: : The original installers of the system setup stty lines in the
: : /etc/rc2.d/S80lp file for each 'printer' port.  Reading the Unix doc,
: : that appears only necessary for direct printing.  I'm always spooling.

: This is un-necessary unless you're sending print jobs directly to
: the port.  I suggest you get rid of these.

Ooh.  I'm not sure of that.  This may depend on OS versions, but I have
definitely seen situations where setting it in the script would not
work.

Besides, it certainly does no harm, and there may come a day when you
*want* to print direct.  I'd leave 'em in.

--
Tony

 
 
 

xon/xoff problem

Post by Bill Vermilli » Thu, 04 Apr 1996 04:00:00




>: : The original installers of the system setup stty lines in the
>: : /etc/rc2.d/S80lp file for each 'printer' port.  Reading the Unix doc,
>: : that appears only necessary for direct printing.  I'm always spooling.

>: This is un-necessary unless you're sending print jobs directly to
>: the port.  I suggest you get rid of these.
>Ooh.  I'm not sure of that.  This may depend on OS versions, but I have
>definitely seen situations where setting it in the script would not
>work.

The {DRAIN} in the current scripts is supposed to cure that - I
say supposed to - as I've had to resort to the old rc2.d
(actually I put them in rc.d/8/userdef) for problems on current
versions of the OS.

Quote:>Besides, it certainly does no harm, and there may come a day when you
>*want* to print direct.  I'd leave 'em in.

Can't hurt - might help.    Just like taking vitamins.

Bill

--

 
 
 

xon/xoff problem

Post by Jeff Lieberma » Fri, 05 Apr 1996 04:00:00


: : This is un-necessary unless you're sending print jobs directly to
: : the port.  I suggest you get rid of these.

: Ooh.  I'm not sure of that.  This may depend on OS versions, but I have
: definitely seen situations where setting it in the script would not
: work.

: Besides, it certainly does no harm, and there may come a day when you
: *want* to print direct.  I'd leave 'em in.

Sure.  Some ancient network printing spooler script sent an
escape sequence that turns OFF all flow control in the printer.
Nobody noticed with the lpr/lpd interface, but don't try useing
that script with a serial port.

There were some other atroscities around that wouldn't flush
their buffers after each job.  The serial port would be closed
before the job was done.  I had to add:
        /usr/spool/lp/bin/drain.output 1
to the end of each script and that headache went away.

I do have customer scribbling directly to a port.
Some of the rationalizations are:
1.  Programmer didn't know how to deal with spooler.
2.  Programmer couldn't figure out how to cancel a
print spooler job so they simply bypassed the spooler.
3.  Programmer couldn't make it work because of bogus
print spooler scripts so they bypassed the spooler.
4.  Medical office management consultant wants to sell
one printer per desk (for real!).
5.  Programmer wants to have print job start instantly.
6.  Programmer duz it just like MSDOS duz it.
and so forth.  You get the idea.  The only real reason I could
see for NOT using the spooler is that the spooler is slow.
This is not a problem with PCL5, but becomes a big problem
with postscribble printers which belch HUGE amount of data.
I ignore the problem completely and usual go to a network based
printer scheme and forget about strange serial or resource hog
parallel interfaces.  Unfortunately, the hpnp spooler script
that comes with EFS110 and appears in two pieces doesn't bother
to pass lp parameters.  Some day when I get ambitious...

[x] Email to author  [ ] To mailing list  [x] Posted to newsgroup
--
# Jeff Liebermann  Liebermann Design  150 Felker St #D  Santa Cruz  CA  95060

# 408.699.0483 digital_pager    73557,2074  cis [don't]

 
 
 

xon/xoff problem

Post by Tony Lawren » Fri, 05 Apr 1996 04:00:00


: and so forth.  You get the idea.  The only real reason I could
: see for NOT using the spooler is that the spooler is slow.

Forms alignment.

--
Tony

 
 
 

xon/xoff problem

Post by Joel Willia » Sat, 06 Apr 1996 04:00:00




>: We have a Fujitsu line printer running at 9600 buad with xon/xoff
>: interface into an Arnet multiport box.  Lately we've been experiencing
>(...)
>: stty 9600 -parenb cs8 ixon ixoff -ixany clocal 0<&1
>That should work.  I would suggest:
>    stty 9600 ixon ixoff -ixany -parenb cs8 clocal onlcr opost 0<&1
>I believe that the opost is required or no post processing occurs.
>: I've eliminated the cable (four wire, flat), connector and Arnet port
>: by using a different port cable and connector with the same strange
>: errors.
>I don't believe it.  Your setup is fine, but it exhibits all
>the classic symptoms of handshake failure.  Here's how you test it:
>    Print a long document to the printer.  /etc/termcap is fine.
>    Install a breakout box (with lights) on the serial port.
>    Stomp on the ON-LINE button randomly.
>    Data flow should stop when the printer is off-line and
>            resume when it is returned on-line.
>My gut feel is that the TD wire (that which sendeth the xon and xoff
>characters back to Arnet) is broken.
>: The old addage 'boot and try again' seems to help as after
>: booting, the printer works fine for a week or two and then things go
>: south.
>I suspect your printer has a large buffer and will work just fine
>for jobs that fit.  My guess is that it only shows up for large jobs.
>: The original installers of the system setup stty lines in the
>: /etc/rc2.d/S80lp file for each 'printer' port.  Reading the Unix doc,
>: that appears only necessary for direct printing.  I'm always spooling.
>This is un-necessary unless you're sending print jobs directly to
>the port.  I suggest you get rid of these.
>[x] Email to author  [ ] To mailing list  [x] Posted to newsgroup
>--
># Jeff Liebermann  Liebermann Design  150 Felker St #D  Santa Cruz  CA  95060

># 408.699.0483 digital_pager    73557,2074  cis [don't]


To follow up, we've isolated the line printer as the culprit here.
There seems to be a problem where the printer is losing characters
that are sent to it.  We changed the printer from serial xon/xoff to
parallel to eliminate handshaking problems and now we're seeing
instances of characters not printing.  I imagine when the printer was
configured for xon/xoff, sometimes the xon or xoff character would be
dropped, hence, 'frozen printer' or buffer overflow.

Thanks for all the advice from my peers.

Joel Williams


MIS Director            http://www.pmso.com
Physicians' MSO