Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/17/84; site ittvax.UUCP
Path: utzoo!linus!decvax!ittvax!long
From: long@ittvax.UUCP (H. Morrow Long [Systems Center])
Newsgroups: net.unix-wizards
Subject: Re: 1-second resolution of process accounting times
Message-ID: <1646@ittvax.UUCP>
Date: Tue, 2-Apr-85 08:15:08 EST
Article-I.D.: ittvax.1646
Posted: Tue Apr  2 08:15:08 1985
Date-Received: Fri, 5-Apr-85 13:07:11 EST
References: <2228@vax4.fluke.UUCP>
Distribution: net
Organization: ITT-ATC, Stratford Ct.
Lines: 27


	We noticed this change in 4.2bsd very quickly after conversion
	from 4.1bsd. We had ported the System III accounting package to
	4.1bsd.  We noticed that under 4.2bsd many of the totals for
	CPU usage were coming up zero for normal users.

	We then were able to attribute it to the coarse time
	measurement being used under 4.2bsd (seconds) and the fact
	that many of the 'average' processes on a system never use at
	least one second of cpu time and therefore show up with zero
	time.

	We modified the accounting routines in sys/kern_acct.c to
	provide utime, stime and etime in 60ths of a second (the 4.1bsd
	way) rather than seconds.

-- 

				H. Morrow Long
				ITT-ATC Systems Center,
				1 Research Drive Shelton, CT  06484
				Phone #: (203)-929-7341 x. 634
	
path = {allegra bunker ctcgrafx dcdvaxb dcdwest ucbvax!decvax duke eosp1
	ittral lbl-csam milford mit-eddie psuvax1 purdue qubix qumix 
	research sii supai tmmnet twg uf-cgrl wxlvax yale}!ittvax!long