pim@ele.tue.nl (Pim Buurman) (03/09/90)
We have here rather big problems with printing. Both lpd/lpr and our own printer-communication programs do/didn't not work well. Our configuration is: 20 apollo's, were running sr9.7 and are updated to 10.2 . Environment: bsd. One Apple Laserwriter+ names ps (for PostScript). Communication with printer: using TranScript software, but already rewritten because of problems with the signals. It now works with forking itself, creating a pipe between parent and child, and if necessary putting a filter between parent and child. Behavior under 9.7: usually good, but sometimes troublesome. Ex & go helps. I will describe the problems we have under sr10.2: 1) The error-log (lf in /etc/printcap) is only logging the errors of the of-filter. The important log of the if-filter (the real communcation program) is logged in /usr/spool/lpd/ps/err???. This log is removed after each job. This would not hurt, if everything was OK. 2) From every node, lpr queues a job, but says: jobs queues, but cannot start daemon. /etc/lpc/ restart ps works, but I am not a daemon. 3) The sio-line was configured wrong. It appeared finally, that an entry in /etc/printcap was wrong !! It did not hurt till now. 4) When a filter was used in the pipe between parent and child (see above) it would finish, but the shell it was started with, did not die. The only remedy was to kill this shell, and everything was allright. But, I am not a killer. Due to these problems, and because the printer is used heavily, we decided to reinstall the sr9.7 lp[dr]. This helps on points 1 and 2, but did not help on point 4 and gave one other problem. To solve problem 4, I have rewritten the filters and the communication program. Now a filter is used always, and the communication program does not fork anymore. An old bug in this program was also found. The new problem was with printing long files. The sio-line was still not configured right: the input_sync was FALSE. Because the printer tells, what it is doing, it should be TRUE. The line was set on TANDEM, but this does not help. We solved this by setting the line's input_sync in /etc/rc.local . We still have the following question: Can we use the 10.2 lp[dr]. Or: can we solve problems 1 and 2 ? Thank you. Pim Buurman. We set with fc/fs in /etc/printcap TANDEM = TRUE, but this does not imply