Cisco Enterprise Print System

Sounds like sockxfer is playing up. When you use lang as Raw, it uses
sockxfer, when you use lang as H it uses jdxfer which is a different
program. We don't use RAW much since we insist on postscript printers.

Frankly this shouldn't be hard to replace. All sockxfer has to do is
open a port 9100 connection, barf down it and then close the
connection.

Damian


Jim Cable wrote:
> 
> I am finally getting some reliable results from a non PostScript HP5Si.
> I set the type to RP ( Printer or machine using LPR protocol ) ,
> language to RAW, and queue to LPT1. Now text and pcl prints work great.
> I was having no end of trouble printing to RAW with a type of JD ( HP
> printer with JetDirect Internal ). Jobs would print but would stick in
> the queue.  Type JD only seems to work well with language H on Color
> LaserJet 4500's.
> 
> --
> Jim Cable
> Senior Developer
> Norfolk Southern Corporation
> 404-582-6478
> ________________________________
> -----==
> ----==   _
> ---==   (_)__  __ ____  __
> --==   / / _ \/ // /\ \/ /
> -=====/_/_//_/\_,_/ /_/\_\
> The choice of a GNU generation
> 
> _______________________________________________
> Ceps-discuss mailing list
> Ceps-discuss@...
> http://lists.sourceforge.net/mailman/listinfo/ceps-discuss

-- 
Damian Ivereigh                   Cisco Systems, Australia
damian@...                  +61 2 8448 7344

                 "Free your mind."

Jim Cable wrote:
> 
> Has there been any experimentation into making the failover process
> automatic?

Not much really. We have been stymied by the problem of network
partitioning. For example when you have two major hub sites each with
a bunch of CEPS servers. If the link between the two dies, then is one
half down or what?

We have also been concerned with how we decide that a server is really
down. It could just be a simple ping test. However if we do more then
we could start getting a cascading failure if a server gets overloaded
and therefore appears to be down and transfers all the load to another
server.

However these are just excuses. We really should do some work on this.

Damian

-- 
Damian Ivereigh                   Cisco Systems, Australia
damian@...                  +61 2 8448 7344

                 "Free your mind."

I believe for ssh to work, the recipient must have the public key of
the calling host in its /etc/ssh_known_hosts.

Damian

ben@... wrote:
> 
> This can be any number of things. I have had the problem many times
> myself. SSH is very particular with things. Usually, I figured it out
> by doing ssh -v and seeing why it throws away the authentication that
> I am trying to do.
> 
> -ben
> 
> > I created .rhosts in root's home on each CEPS box so that I dont need to
> > put in a password to be authenticated via rsh. I'm trying to do the same
> > for ssh but I cant seem to get rid of the password prompt. Has to be
> > something simple. BTW I changed allmkprint back to using rsh and it
> > works beautifully.
> >
> > --
> > Jim Cable
> > Senior Developer
> > Norfolk Southern Corporation
> > 404-582-6478
> > ________________________________
> > -----==
> > ----==   _
> > ---==   (_)__  __ ____  __
> > --==   / / _ \/ // /\ \/ /
> > -=====/_/_//_/\_,_/ /_/\_\
> > The choice of a GNU generation
> >
> > _______________________________________________
> > Ceps-discuss mailing list
> > Ceps-discuss@...
> > http://lists.sourceforge.net/mailman/listinfo/ceps-discuss
> 
> _______________________________________________
> Ceps-discuss mailing list
> Ceps-discuss@...
> http://lists.sourceforge.net/mailman/listinfo/ceps-discuss

-- 
Damian Ivereigh                   Cisco Systems, Australia
damian@...                  +61 2 8448 7344

                 "Free your mind."

When you use raw mode, then a rather grotty program (I wrote) called
sockxfer gets used to send the job to the printer on port 9100. It
sounds like it is hanging for some reason. Drop a file called .debug
into the spool dir and put a number like 10 into it. It will put
sockxfer into debug mode and it will create a file called log.<pid>

See if you can see anything happening in there.

I am in the process of replacing this prog with one written in perl
that will be much more efficient.

Damian

Jim Cable wrote:
> 
> Why do RAW prints to this HP5SI fail to spool? The first job prints and
> then it just sits in queue and never gets removed. If you delete that
> job then the next job prints, etc..
> 
> sddb entry follows:
> > device "p125-08n-5si" show
> Ownid/Seqno     : GAATL/42
> Last mod by     : admin
> Last mod time   : Thu Jan 20 15:57:24 2000
> Name            : p125-08n-5si
> Alias(es)       : p125-08n-5si
> Active          : Y
> Type            : JD - HP printer with JetDirect Internal
> Location Code   : GAATL-125-06N - CYO
> Location        : 125 Spring, 8th Floor N
> Contact         : Jim Cable
> Email Contact   : jscable@...
> Memory (KB)     : 4096
> All LP Options  : psize=letter
> Model (auto)    : laserjet 5si - HP LaserJet 5si
> Language        : R - Raw (no changes)
> Duplexer        : Y
> Browseable      : Y
> PC Printable    : Y
> Exclusive       : Y
> Auto Delete     : Y
> Daily Check     : Y
> PCL Printing    : Y
> Critical        : N
> Hostname/IP     : p125-08n-5si (167.121.122.203 - dynamic)
> H/W Addr        : 0060b00a54e0
> 
> Setting the language to H works fine but I was testing the ability to
> pass prints through CEPS without any changes. Is this a conflict with
> the type/language selections?
> 
> --
> Jim Cable
> Senior Developer
> Norfolk Southern Corporation
> 404-582-6478
> ________________________________
> -----==
> ----==   _
> ---==   (_)__  __ ____  __
> --==   / / _ \/ // /\ \/ /
> -=====/_/_//_/\_,_/ /_/\_\
> The choice of a GNU generation
> 
> _______________________________________________
> Ceps-discuss mailing list
> Ceps-discuss@...
> http://lists.sourceforge.net/mailman/listinfo/ceps-discuss

-- 
Damian Ivereigh                   Cisco Systems, Australia
damian@...                  +61 2 8448 7344

                 "Free your mind."

Well, I dont know about using the token ring, but the ones that have
ethernet cards, we simply use the JD type.  They boot close enough to the
HP's that there is no reason for a different type.

-Adam

On Mon, 17 Jan 2000, John Housden wrote:

| I am trying to set up a Lexmark Optra R+ printer with internal (token ring) network card.
| 
| When using 'pradmin', I see there is a device type for Lexmark but it says '(do not use)'.
| Is this something still on the to-do list, or has work on this been abandoned?
| 
| I have tried setting up the printer as both LX for Lexmark and JD for HP jetdirect, but neither prints. The job queues up OK, but disappears into the ether after that!
| 
| Any tips (apart from but HP printers!) ?
| 
| Regards,
| 
| John
| 
| ------
| John Housden <john.housden@...>
| Unix Support Analyst
| TRW Information Services Centre
| Old Church Road, Clevedon, North Somerset, BS21 6NJ.
| Tel: +44 (0)1275 512045
| 
| 
| 
| _______________________________________________
| Ceps-discuss mailing list
| Ceps-discuss@...
| http://lists.sourceforge.net/mailman/listinfo/ceps-discuss
|

In the early days of Lexmarks, we used a different type, which allowed
us to use a different filter script. However the more modern ones are
similar enough to an HP printer that we can just use an HP type (JD).
So we stopped keeping the lexmark filter script up to date.

All the Optra series will work just fine with type JD. So as for why
it is disappearing after that I am not sure. You can try talking
direct to the printer and see if that works thus:-

cat <job> | /usr/local/ciscolp/bin/jdxfer <Ip address>

If that works then the filter script is fowling up, if it doesn't
jdxfer's your problem. Let us know and we'll help you diagnose
further.

Damian

Adam Crews wrote:
> 
> Well, I dont know about using the token ring, but the ones that have
> ethernet cards, we simply use the JD type.  They boot close enough to the
> HP's that there is no reason for a different type.
> 
> -Adam
> 
> On Mon, 17 Jan 2000, John Housden wrote:
> 
> | I am trying to set up a Lexmark Optra R+ printer with internal (token ring) network card.
> |
> | When using 'pradmin', I see there is a device type for Lexmark but it says '(do not use)'.
> | Is this something still on the to-do list, or has work on this been abandoned?
> |
> | I have tried setting up the printer as both LX for Lexmark and JD for HP jetdirect, but neither prints. The job queues up OK, but disappears into the ether after that!
> |
> | Any tips (apart from but HP printers!) ?
> |
> | Regards,
> |
> | John
> |
> | ------
> | John Housden <john.housden@...>
> | Unix Support Analyst
> | TRW Information Services Centre
> | Old Church Road, Clevedon, North Somerset, BS21 6NJ.
> | Tel: +44 (0)1275 512045
> |
> |
> |
> | _______________________________________________
> | Ceps-discuss mailing list
> | Ceps-discuss@...
> | http://lists.sourceforge.net/mailman/listinfo/ceps-discuss
> |
> 
> _______________________________________________
> Ceps-discuss mailing list
> Ceps-discuss@...
> http://lists.sourceforge.net/mailman/listinfo/ceps-discuss

-- 
Damian Ivereigh                   Cisco Systems, Australia
damian@...                  +61 2 8448 7344

                 "Free your mind."

mkhosts is simply a shell script that reads the pserver and main tables
in sddb, so if your getting a blank /etc/hosts at the end, sddb must
have failed (it should catch this better I guess). There are several
reason why this could be happening. You will have to give me more
details - I expect that the sddb server is not starting because the
print server can't find itself in the pserver table (did you change the
name of the server too)?

sddb used the IP address as a key to figure out which server is which.
Not ideal I agree, but I have yet to change it. So when you change the
Ip address of a master server, the sddb server will come up and think
that some other machine is the master (with a different IP address). So
the sddb server drops into slave mode and waits for the master to assert
itself. Since there now is no master server, this will never happen.
There is currently no election process for a new master (I was working
on it, but got diverted).

The solution is to manually tell sddb that it is the master. Once the
sddb server is up, run up the sddbd command and use the command master.
Thus

sddbc -c localhost master

That will force the sddb server to be a master and all will be well
again.

Damian


Jim Cable wrote:
> 
> Thanks for the help on allmkprint.
> 
> I'm having unusual symptoms when I change the IP address of a CEPS box.
> Strange things like mkhosts generates a blank list of servers or sddbd
> wont let you change masters. Is there a hidden table that needs to be
> modified?
> 
> --
> Jim Cable
> Senior Developer
> Norfolk Southern Corporation
> 404-582-6478
> ________________________________
> -----==
> ----==   _
> ---==   (_)__  __ ____  __
> --==   / / _ \/ // /\ \/ /
> -=====/_/_//_/\_,_/ /_/\_\
> The choice of a GNU generation
> 
> _______________________________________________
> Ceps-discuss mailing list
> Ceps-discuss@...
> http://lists.sourceforge.net/mailman/listinfo/ceps-discuss

out of the box allmkprint will rsh to each machine, so you just need to
have a good .rhosts on the remote machines.  We run the allmkprint as root
and root from the central server is allowed rsh access to all the other
servers.

We havent completely gone to ssh yet because I havent sat down and set up
a ssh key distribution yet.  With all the servers that we have, we need to
make sure all the public keys are on all the servers.

As for the load.... we had 300 printers on a P2 w/256 mb ram before we had
problems.  Now we have backed down to about 100-200 printers on a P3
w256mb ram.

The biggest factor that we have found is the ram.  Justin and I are 1/2
working on figuring out some true way to actually determine the resources
needed for a printer, so we can figure out some intelligent capacity
plans.

-Adam

On Thu, 13 Jan 2000, Jim Cable wrote:

| I have a CEPS development system up and running using four machines and
| four printers. How do I setup/use allmkprint? Is there someplace that I
| have to put in a username/passwd for allmkprint to use with ssh?  I'm
| getting tired of manual login/mkprint. Heh.
| 
| Are there any other critical CEPS components that need to be working
| before I go into pre-production testing? One of our departments wants
| CEPS in a bad way. They have been having major print problems with NT
| for several weeks and heard about my CEPS project. I built two CEPS
| boxes for them yesterday. These boxes are our standard Gateway PII 350
| desktops with 128mb ram. The load will be 7 print queues and 110 NT
| clients per box.Probably overkill but these users print alot of large
| print jobs and until I gain confidence in system performance, I want
| some horsepower.
| 
| The goal is to have everything in place for testing on Monday. I will
| probably setup two dns aliases and have the clients print to that.
| 
|

Actually Adam that's not true we are using ssh for allmkprint. However
the key problem you mention is real. If understand ssh correctly, we
could just put the public keys for the central servers (i.e. the
trusted ones) into the dist and then we would have the access. Just we
couldn't be sure that we are indeed contacting the machine we specify
- which probably doesn't matter anyway.

Damian

Adam Crews wrote:
> 
> out of the box allmkprint will rsh to each machine, so you just need to
> have a good .rhosts on the remote machines.  We run the allmkprint as root
> and root from the central server is allowed rsh access to all the other
> servers.
> 
> We havent completely gone to ssh yet because I havent sat down and set up
> a ssh key distribution yet.  With all the servers that we have, we need to
> make sure all the public keys are on all the servers.
> 
> As for the load.... we had 300 printers on a P2 w/256 mb ram before we had
> problems.  Now we have backed down to about 100-200 printers on a P3
> w256mb ram.
> 
> The biggest factor that we have found is the ram.  Justin and I are 1/2
> working on figuring out some true way to actually determine the resources
> needed for a printer, so we can figure out some intelligent capacity
> plans.
> 
> -Adam
> 
> On Thu, 13 Jan 2000, Jim Cable wrote:
> 
> | I have a CEPS development system up and running using four machines and
> | four printers. How do I setup/use allmkprint? Is there someplace that I
> | have to put in a username/passwd for allmkprint to use with ssh?  I'm
> | getting tired of manual login/mkprint. Heh.
> |
> | Are there any other critical CEPS components that need to be working
> | before I go into pre-production testing? One of our departments wants
> | CEPS in a bad way. They have been having major print problems with NT
> | for several weeks and heard about my CEPS project. I built two CEPS
> | boxes for them yesterday. These boxes are our standard Gateway PII 350
> | desktops with 128mb ram. The load will be 7 print queues and 110 NT
> | clients per box.Probably overkill but these users print alot of large
> | print jobs and until I gain confidence in system performance, I want
> | some horsepower.
> |
> | The goal is to have everything in place for testing on Monday. I will
> | probably setup two dns aliases and have the clients print to that.
> |
> |
> 
> _______________________________________________
> Ceps-discuss mailing list
> Ceps-discuss@...
> http://lists.sourceforge.net/mailman/listinfo/ceps-discuss

-- 
Damian Ivereigh                   Cisco Systems, Australia
damian@...                  +61 2 8448 7344

                 "Free your mind."

Ahh.. we have ssh configured so that it allows rhost authentication.  
This wont enforce all the key exchange stuff to the full extent.

Eventually we should get to a point where the root account must present a
valid user identity, and do .shost authentation only.  I'll add that to my
todo list.

-Adam

On Fri, 14 Jan 2000, Damian Ivereigh wrote:

| Actually Adam that's not true we are using ssh for allmkprint. However
| the key problem you mention is real. If understand ssh correctly, we
| could just put the public keys for the central servers (i.e. the
| trusted ones) into the dist and then we would have the access. Just we
| couldn't be sure that we are indeed contacting the machine we specify
| - which probably doesn't matter anyway.
| 
| Damian
| 
| Adam Crews wrote:
| > 
| > out of the box allmkprint will rsh to each machine, so you just need to
| > have a good .rhosts on the remote machines.  We run the allmkprint as root
| > and root from the central server is allowed rsh access to all the other
| > servers.
| > 
| > We havent completely gone to ssh yet because I havent sat down and set up
| > a ssh key distribution yet.  With all the servers that we have, we need to
| > make sure all the public keys are on all the servers.
| > 
| > As for the load.... we had 300 printers on a P2 w/256 mb ram before we had
| > problems.  Now we have backed down to about 100-200 printers on a P3
| > w256mb ram.
| > 
| > The biggest factor that we have found is the ram.  Justin and I are 1/2
| > working on figuring out some true way to actually determine the resources
| > needed for a printer, so we can figure out some intelligent capacity
| > plans.
| > 
| > -Adam
| > 
| > On Thu, 13 Jan 2000, Jim Cable wrote:
| > 
| > | I have a CEPS development system up and running using four machines and
| > | four printers. How do I setup/use allmkprint? Is there someplace that I
| > | have to put in a username/passwd for allmkprint to use with ssh?  I'm
| > | getting tired of manual login/mkprint. Heh.
| > |
| > | Are there any other critical CEPS components that need to be working
| > | before I go into pre-production testing? One of our departments wants
| > | CEPS in a bad way. They have been having major print problems with NT
| > | for several weeks and heard about my CEPS project. I built two CEPS
| > | boxes for them yesterday. These boxes are our standard Gateway PII 350
| > | desktops with 128mb ram. The load will be 7 print queues and 110 NT
| > | clients per box.Probably overkill but these users print alot of large
| > | print jobs and until I gain confidence in system performance, I want
| > | some horsepower.
| > |
| > | The goal is to have everything in place for testing on Monday. I will
| > | probably setup two dns aliases and have the clients print to that.
| > |
| > |
| > 
| > _______________________________________________
| > Ceps-discuss mailing list
| > Ceps-discuss@...
| > http://lists.sourceforge.net/mailman/listinfo/ceps-discuss
| 
|

Partly. We currently have driver download capability for Win9x, but
not for NT and Win2k. We are dependant on Samba providing this
capability, which by all accounts is very shortly going to be a
available.

Damian

Rajeev Agrawala wrote:
> 
> Hi,
> 
> I am planning to tryout CEPS for printing here. I wanted to find out if
> CEPS supports driver downloads to NT/W2k/Win9x.
> 
> I haven't looked at the documentation very thouroughly, and pl. excuse
> me if this is mentioned there.
> 
> Thanks,
> 
> rajeev
> 
> _______________________________________________
> Ceps-discuss mailing list
> Ceps-discuss@...
> http://lists.sourceforge.net/mailman/listinfo/ceps-discuss

-- 
Damian Ivereigh                   Cisco Systems, Australia
damian@...                  +61 2 8448 7344

                 "Free your mind."