Path: utzoo!utgpu!news-server.csri.toronto.edu!mailrus!cs.utexas.edu!usc!snorkelwacker!mit-eddie!uw-beaver!fluke!gtisqr!stu
From: stu@gtisqr.uucp (Stu Donaldson)
Newsgroups: comp.unix.i386
Subject: Re: lpadmin(8) question
Summary: safer form of lock.
Keywords: lp systemv
Message-ID: <1990Jun6.200318.1619@gtisqr.uucp>
Date: 6 Jun 90 20:03:18 GMT
References: <453@van-bc.UUCP> <162@twg.bc.ca>
Reply-To: stu@gtisqr.UUCP (Stu Donaldson)
Organization: Global Tech Int'l Inc.
Lines: 53

In article <162@twg.bc.ca> bill@.UUCP (Bill Irwin) writes:
>#
>if [ -f /tmp/computer.lock ]
>then
>        while [ -f /tmp/computer.lock ]
>        do
>                sleep 60
>        done
>fi
>touch /tmp/computer.lock

	... model/interface stuff deleted ...

>rm /tmp/computer.lock

>The  only drawback with this approach that I have encountered is when you
>cancel  a print job the lock is not removed.  You have to remember to "rm
>/tmp/computer.lock" after your cancel, otherwise you next jobs will never
>print.

A safer solution that I have used for locks, that solvs the problem of
the lock file being left around is:

while [ -f /tmp/computer.lock ]
do
	if ps -p`cat /tmp/computer.lock` >/dev/null 2>&1
	then
		sleep 60
	else
		rm -f /tmp/computer.lock	# lock invalid, so remove.
	fi
fi

echo $$ >/tmp/computer.lock		# I should lock it.

# for added security, the following test can be made.

sleep 2
if [ "$$" != "`cat /tmp/computer.lock`" ]
then
	... lock failed, go back and try again? ...
fi

>-- 
>Bill Irwin - TWG The Westrheim Group - Vancouver, BC, Canada
>~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>uunet!van-bc!twg!bill     (604) 431-9600 (voice) |     UNIX Systems
>Bill.Irwin@twg.bc.ca      (604) 431-4629 (fax)   |     Integration
-- 
Stu Donaldson          UUCP: {smart-host}!gtisqr!stu
Engineering Manager    ARPA: gtisqr!stu@yang.cpac.washington.edu
Global Technology      Bell: (206) 742-9111
Mukilteo, Washington.