Q:
Topic thread:
Printer Systems (PRINT - NA/ATS)
PSF/AIX
Customer has 2 3160-001 and 2 3935-001 printers attached to the network
and driven by PSF/AIX, all in the same room. We are in the process of
trying to get them to upgrade their printers and to IPM. They said they
thought they should keep one of the 3935s because it was "faster" on
multiple small jobs. They conducted a test as follows. Set-up 20 one
page jobs for each printer and release the que. The 3935 takes the
first job runs continuously until all jobs in que are printed. The 3160
take the first job and begins, prints 4 jobs, pauses, prints 4 jobs,
pauses, etc until the que is empty. The net result is the 3935 finishes
the 20 one page jobs before the 3160.
Do you think this could be a function of PSF/AIX and how the printers
are defined? OR Is it a function of how the printer's control unit
processes the jobs (they both have AFCCU control units)? Any
suggestions?
Please advise if further info is needed.
A:
It sounds to me like they need to define additional queue devices in
PSF/AIX for the 3160. The default is four, which means that PSF/AIX
is running four different ainbe processes, one of which can actually
print while the other three are preprocessing/transforming/converting
jobs in preparation for printing. A queue device doesn't free up until
the last page of the job it's handling reaches the output stacker of
the printer. It would be enlightening to know what the queue status
looks like while the jobs are processing (qchk -Pprtname -w5 will
update the status every five seconds as long as there are jobs in
the queue) to understand what's really going on.
It could be that the 3935s have more queue devices defined, or could
have a shorter paper path (?) so that jobs reach the stacker more
quickly -- not sure exactly why the difference between the two. I can
consult with someone more familiar with the hardware if you'd like?
One difference is that the 3160-001 is 240-pel, while the 3935 is
300-pel. Is there any image in the one page documents? What is the
input data stream? I'm just trying to think of things that might
take additional AFCCU time to process... There may be some difference
in how the two printers handle the inter-job processing, but I'd have
to defer to a hardware specialist again.
There are also PSF/AIX tuning parameters that can be set to retain
different numbers of AFP resources at the printer between jobs. It
would be worthwhile to compare their profiles to see if there are any
PSF settings that might account for the difference. To capture a
readable version of the PSF printer profile:
cd /var/psf/
mkprof -v .profile > .mkprof
For example, if my PSF/AIX queue name is 3160q, then:
cd /var/psf/3160q
mkprof -v 3160q.profile > 3160q.mkprof
If you want to add additional queue devices to the PSF/AIX queue for
the 3160, you can do so through smitty mkpqprt. You would select
a backend type of "other -- User Defined Backend", then on the
following panel:
Add a Queue Device to an Existing Queue
Type or select values in entry fields.
Press Enter AFTER making all desired changes.
(Entry Fields)
* Name of DEVICE to add ()
* QUEUE to which to attach device ()
* BACKEND PROGRAM pathname ()
BACKEND OUTPUT FILE pathname ()
ACCESS MODE of backend output file write only
Print HEADER pages? never
Print TRAILER pages? never
ALIGN page if printer has been idle? no
Number of FORM FEEDS when printer goes idle (0)
For "Name of DEVICE to add", follow your existing naming conventions
to add a fifth or subsequent qdev (probably dev5, dev6, etc..
For "QUEUE to which attach device", enter the PSF/AIX printer name for
the 3160-001.
For "BACKEND PROGRAM pathname", use the exact information that is
already in the /etc/qconfig file for the other queue devices on this
queue -- something along the lines of "/usr/lpp/psf/bin/ainbe 3160q"
That's all I can think of from the software side. If you would like
me to transfer this to a hardware specialist who will know more about
the internal differences in the 3160 and 3935, or if you have any
additional PSF/AIX-specific questions, please reopen this item.
Thanks for using ViewBlue.
Q:
Here's some feed back from the customer. Stay tuned.
Bill,
Sorry I haven't got back to you sooner. Our IBM printer CE has
captured a trace and was working their end also. We are going to
work
with him as well and see what they come up with.
So far we haven't changed anything.
Thanks for the info. and I will let you know what unfolds.
Ken
A:
Thanks for the update. I'm sure L2 hardware will get L2 software
involved if necessary (if they aren't already). If I can do anything
else, let me know.
Q:
Kristin, Just a quick question. What do you think about the attached in
regard to the subnet and gateway addresses may be causing the delay? Not
sure what the ip addressing is in the 3935 or if it would be relevant.
Your thoughts, Thanks.
Following from CE's support:
Mickey, here is the result of trace analysis by Charles Lujan;
Trace does show that 4 pages are stacked within a second, then a delay
of 47 seconds, and 4 more pages are stacked. However, it also shows that
the pages are coming into the printer slowly. For those 8 pages, it
takes 48 seconds, much too slow to be able to print at 60 ppm. The IPDS
trace was not taken, so I can't see what is on each page. I don't
suspect that they are overly complex since we can rasterize them in
about a second per page.
I suspect that they have a networking problem. In looking at the config
page, it looks like the IP address is 125.3.0.214, and the gateway
address is 125.0.0.2, and the subnet is 255.0.0.0. Typically, one would
want the gateway on the same subnet ( on the 125.3.x.x) and a mask of
255.255.252.0. In this way, you limit the number of addresses that have
to be searched to find the printer. In other words, there could be a lot
of extraneous traffic on the printers subnet. I don't see any other
errors.
I would suggest that the CE explain that the customer needs to look at
his network.
A:
I agree that those are very odd gateway and subnet mask addresses.
I would also be curious to see what the same values are on the RS/6000.
Please trying the values suggested. If this doesn't help, please
respond with those values for the RS/6000 and I will try and determine
if that should affect the performance or not.
Thank you.
S e a r c h - k e y w o r d s:
3160 performance psf/aix psf aix 3935 subnet mask gateway tcpip ip
address ipaddress
WWQA: ITEM: RTA000159419 ITEM: RTA000159419
Dated: 10/1999 Category: XPSF6000
This HTML file was generated 2000/11/30~13:34:11
Comments or suggestions?
Contact us