tty crashing after certain amount of data

[Michael Campbell]

Folks

We have a printer directly connected to one of our servers. After roughly 50 pages however, it seems to lose it's settings and the line-up etc goes wrong. This problem has only come up since we switched this printer from /dev/tty2a2 to /dev/tty2a1. Does anyone have any ideas on this?

Any Help Appreciated

Michael

[Jordan Bean]

Is the interface script invoking stty against the device file? For example, the canned model scripts use:

stty raw 9600 -parenb cs8 ixon -istrip clocal

Let's see what these have to say:

stty -a < /dev/tty2a1
stty -a < /dev/tty2a2

[Darren Prior]

hi Michael,

I'd compare the 2 device files using lssf, also check if you have an entry in inittab that uses stty to set up the port (baud rate, handshaking, etc)

regards,

Darren.

[Michael Campbell]

Folks

The stty -a settings for tty2a1 are the same as what they were for tty2a2 when it was working (when I run stty -a The lssf output looks fine for both.
Any idea what else I can look at?

Regards

Michael

[Michael Campbell]

Darren

There is an entry in the inittab for the stty settings which we edited just to change the device name.

Rgds

Michael

[Darren Prior]

Hi Michael,

The inittab change should be OK if you just changed the device file over - did you use init q to re-read inittab?

Without knowing the server model and the type of mux you are using it is hard to give you a definitive answer. Is there any reason for you moving from one port to the other? Did it require any change in cabling? Is there anything port specific in the model/interface scripts that sets the port?

The fact that you cannot use stty -a on the port suggests that it could be hung - it may be worth shutting the server down and powercycling it in order to clear it.

Check to see if anything else is trying to use the port by running fuser /dev/*2a1 and checking out any PIDs that are returned.

regards,

Darren.

[Michael Campbell]

Darren

It is a 9000 N-Class server running 11.0 . It is our production box and so we can't shut it down to try to re-enable tty2a2, this is why we had to swap to tty2a1 after tty2a2 hung.
There were no physical changes except switching to the other mux port and the only code change was to the job which uses the printer and to the inittab.
The lssf output is:
pci_mux0 card instance 2 port a1 hardwired at address 0/8/0/0 /dev/tty2a1

pci_mux0 card instance 2 port a2 hardwired at address 0/8/0/0 /dev/tty2a2

When I run fuser -fu on /dev/tty2a1 I see two processes:
nohup sleep 2000000000
and a "getty" process both of which look normal.

Does any of this give you any more clues?

Regards

Michael

[Darren Prior]

Hi Michael,

I'd say that you really don't want the getty there. :)

Set the getty entry for that port to 'off' rather than 'respawn' and run init q to remove the getty.

Your printer probably isn't very happy about receiving a login prompt whenever it tries to do some handshaking!

regards,

Darren.

[Michael Campbell]

Darren

We have just discovered an istrip parameter missing from tty2a1. I changed this and the "min" value and tested a large file. This seems to work OK now.

Thanks for your help

Michael