Date: Mon, 4 Oct 1999 20:13:36 -0700 (PDT) From: Matthew Dillon <dillon@apollo.backplane.com> To: "Rodney W. Grimes" <freebsd@gndrsh.dnsmgr.net> Cc: grog@lemis.com (Greg Lehey), des@flood.ping.uio.no (Dag-Erling Smorgrav), thyerm@camtech.net.au, ru@FreeBSD.ORG (Ruslan Ermilov), cvs-committers@FreeBSD.ORG, cvs-all@FreeBSD.ORG Subject: Re: top with < 1 second delay (was: cvs commit: src/bin/sleep sleep.1 sleep.c) Message-ID: <199910050313.UAA73339@apollo.backplane.com> References: <199910050258.TAA38780@gndrsh.dnsmgr.net>
next in thread | previous in thread | raw e-mail | index | archive | help
:
:I don't see a need for any such restrictions, a user can also type
:while true; do done and achive about the same thing as top -s0.01.
:
:Presently you can run top -s0, it doesn't kill a system, it just
:drives la up to 1. And even this old slow doggy P100/16MB box
:running 5 xterms, gated, xntpd is still usable for typing this
:email message while top -s0 is running :-).
:
:--
:Rod Grimes - KD7CAX - (RWG25) rgrimes@gndrsh.dnsmgr.net
For anyone who's ever run a multi-user system with lots of users, such
restrictions are often very important. Simply put, users can get into
bad habits very easily. At BEST we had to disable top entirely because
so many users were leaving it running - it wastes a huge amount of cpu
on heavily loaded systems. Yet we did not have to disable the 'pigs'
function in systat. In fact, we had a similar problem with 'ps' --
literally a hundred or more /bin/ps's were being run simultaniously at
the top of the hour from cron and often at other integral times as well.
In the case of /bin/ps, the real problem was that it was forcing all
the processes to swap in trying to read their uarea by default. This led
to my changing the default behavior for ps on swapped out processes
and adding the 'f' option (root-only) to revert to the old behavior.
The issue here is not whether a user can purposefully eat up all the cpu,
but whether he can do it accidently without really knowing what he is
doing. while(TRUE) is a very purposeful action. You cannot equate the
two. It just so happens that people who use 'top' a lot tend to use the
's<VALUE><RETURN>' sequence a lot too (as we found at BEST), and this
is why allowing sub-second specifications by users may be bad for top.
-Matt
Matthew Dillon
<dillon@backplane.com>
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe cvs-all" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199910050313.UAA73339>