Patches
-------
Ensure that your server has been patched with the latest Maintenance
Pack available from:
ftp://ftp.sco.com/pub/unixware7/uw711pk/
and:
ftp://ftp.sco.com/pub/openunix8/ou800pk/
Network Drivers
---------------
Ensure that your server is using the latest network "nd" driver
available from:
ftp://ftp.sco.com/pub/unixware7/drivers/network/
and:
ftp://ftp.sco.com/pub/openunix8/drivers/network/
TCP/IP Tuning
-------------
Ensure that your server has been tuned for TCP/IP. The details are
available from:
Technical Article 113164 for UnixWare 7.1.1
and:
Technical Article 114966 for OpenUNIX 8.0.0/7.1.3
Kernel Tuning
-------------
Ensure that your server's kernel has been tuned in accordance with:
Technical Article 117343, "I am getting abnormal process termination error messages
in lpNet. "
This problem has been reported to SCO Engineering and will be fixed
in a later release of the Maintenance Packs for UnixWare 7.1.1
(currently Maintenance Pack 2) and for OpenUNIX 8.0.0 (Currently
Maintenance Pack 4).
NOTE:
Should you experience problems with HP JetDirect printers giving 2
minute delays it is suggested to set the "retry" field to "0" in
/etc/lp/Systems and stop and re-start the lp scheduler as detailed
above.
After a print job is successfully sent to a remote machine, lpsched
schedules a status request to be sent to this printer after 2
minutes. lpsched, then after 2 mins, requests a status request. If
it finds the printer system busy, for example, if another print
request was in the process of being sent, then the print status
request waits for another 2 mins. If the printer is busy, the
jobids even though printed, never get recycled.
What lpsched needs to do is to add the status request to the
queue of requests to be sent to the printer. Since status requests
have higher priority over print requests, next time the status
request should be sent and if printed then removed from the queue.
This fix will also reduce the likelihood of the print queues filling
up, because querying the status with the job will ensure that the
job is removed more rapidly from the queue.
Should the "lpsched" process die, due to intense printing activity
then a workaround to ensure that "lpsched" is restarted would be to
run a script constantly checking the "lpsched" process in the
"cron":
# cat /local/bin/start_lpsched
:
NUMBER=`ps -ef |grep lpsched|grep -v start_lpsched|grep -v grep|wc
-l`
if [ $NUMBER -le 0 ]
then
echo "lpsched died - `date` " >> /var/lp/logs/lpsched.`date
+'%y%h%d'`
exec /usr/lib/lpsched -d -s >> /var/lp/logs/lpsched.`date
+'%y%h%d'` 2>&1 &
fi
=====================================
Here is the crontab entry:
=====================================
* * * * * /local/bin/start_lpsched
=====================================
Should you experience "lpsched" hanging with during intense printing
activity then "lpsched" should be killed using the "-11" option
which should produce a "core" file in the directory where "lpsched"
was started and it should also be present in /usr/lib/lp.
Should you experience "lpd" printers not printing, "lpstat -t"
or "who -b" and "who -r" hanging, check that /etc/utmpx has not been
removed.
SEE ALSO:
Technical Article 117343, "I am getting abnormal process termination error messages
in lpNet."
Technical Article 114966, "Open UNIX 8, My application returns IP SOCKET ACCEPT
failed [errno 63] when I try to open hundreds of sockets per second."
Technical Article 113164, "My application returns SOCKET accept failed [errno 126]
when I try to open over 200 sockets every second."
Technical Article 111121, "How do I configure my UnixWare 7 machine to print to a
Hewlett-Packard network printer?"
|