Digi c/con16 and Digi AccelePort - Enable/Disable

[JvL45]

About 3 months ago, I got a call from our call center that everyone had been kicked off their terminals. A quick look under /dev revealed that all the tty's (all my digi serial devices) for the call center had been re-inited! This occurred at the same time that the office manager was doing a disable/enable on a bummed out dumb terminal (I have a script that runs asroot to do this). I have 4 more cases in the last 3 months - all occurring when a enable/disable was done.

I have been able to re-create this by the following. 1. Running this script (see below) 2. Putting the system under a heavy, heavy, heavy load. Doing just #1 does nothing. I let it run like this for two days and I couldn't recreate it.

Note - the script is to just provoke the situation. It is not the script we are running to do the enable/disable.

Script:
l /dev/tty[a-d]*
while [ 0 ]
do
date
disable ttyc01
sleep 2
enable ttyc01
sleep 2
done

The equipement used are Digi c/con16 concentrators, 3 Digi AccelePort Host controllers and many wy60 dumb terminals.

I have recreated it on two 5.0.6 servers and I am trying to re-create on a 5.0.4 server. The hardware is a Dell PowerEdge 6600 with 4 gbs of ram, quad 2.5 ghz Xeon processors.

 

[stanhubble]

do not use 5.0.4 on a Xeon system only 5.0.6a and later.

could you post the script you are using to enable/disable?

 

[JvL45]

Thank you for the reply.

On your first point - all of our 5.0.4 servers are on PII's and low end PIII's (ie: 550 mhz)

On your request:

Here is the script. As mentioned, I have it under /tcb/files/rootcmds. Remember, this script below you is not used when I do my test. It is the one above.

if [ "$2" != "" ]
then
echo "\nKilling $1"
ps -t $1 >>/u/httpd_docs/field/reset.log
for i in $2
do
kill -9 $i 2>>/u/httpd_docs/field/reset.log
sleep 2
echo "kill -9 $i"
echo "Killing Process $i for TTY $1" >>/u/httpd_docs/field/reset.log
done

sleep 5
fi;

echo "\nDisabling $1"
echo "Disabling $1">>/u/httpd_docs/field/reset.log
disable $1 2>>/u/httpd_docs/field/reset.log sleep 5

echo "\nEnabling $1"
echo "Enabling $1">>/u/httpd_docs/field/reset.log
enable $1 2>>/u/httpd_docs/field/reset.log sleep 5

ps -t $1 >>/u/httpd_docs/field/reset.log

echo "Remove TMPWRK files"
rm -r /u/tmp/dev/$1 2>/dev/null;

You won't find anything here. But boy, do I have something to add to this string!

On friday I ran my test but disabled 3 of the 4 processors (using cpuonoff). Within minutes, I couldn't log into the system. In a nutshell, /etc ,/dev and part of /usr had been deleted! We'll, I spent the next 1.5 hours restoring the root filesystem! I believe the issue is bigger than configuration or a script. I believe the issue is related to a SCO/P4 issue.