TEMNGT23@ysub.ysu.edu (Lou Anschuetz) (11/21/90)
We have posted some items about this problem before, but thought we might add some additional information. We have now installed 1.01.01 and the patches that have virtually eliminated problems with HPSIO buffer overflow. We continue to have problems with Getty's that won't die from time to time, and the only solution there continues to be a reboot. That is inconvenient since I am having to reboot the Tower 32/700 four times per day -so much for family life, weekends and vacations :-( We also have the strange additional problem that occasionally a person will log off and the getty will clear, but the sh script does not die. The next person logging into that port becomes the previous user with all rights and privileges. This is happening at least once per day. Since my users are invoking a shell script, their exit is via a "kill -9 0" if voluntary (I stole this from an NCR script). If I force them off I do a "kill -1 mainpid; kill -9 anyotherpids". In general a voluntary logoff works correctly. The main problem is people who hang up without going through logoff. These sessions almost always don't die. Not as problematic, but clearly along the same line, folks coming in through the ethernet link who do a telnet quit, but not an exit, will leave certain activities behind. For example, I call pg from the shell script. If the ethernet user quits while the pg is present, the pg will hang around until a reboot. This is not a serious problem, just lots of useless unkillable pids. Of course it is obvious from the above discussion that kill -1 and kill -9 by root have absolutely no effect. Anyone who has any help for the problems will probably contribute mightily to the salvation of my marriage. ;-) PS: an errno 11 on a modem port can make your machine go braindead(!) I've had it happen twice now, both times with extensive corruption of my /etc/passwd file (the one that is hardest to recover of course). Any ideas on that one are also welcome. Thanks in advance Lou Anschuetz root@yfn.ysu.edu temngt23@ysub.ysu.edu temngt23@ysub.bitnet PPS: Thanks to all who sent me vt52 termcap info - there were some folks whose address my machine would not let me return mail to. Thanks in any case!