fischer@netmbx.UUCP (Axel Fischer) (06/08/89)
I'm just wondering why my system *always* displays a load average of 1.00 when I do nothing. It's Xenix Sys V 386 Release 2.3.1 When i execute the /usr/bin/w command and nothing has happend the last 15 minutes (excludes gettys, a talk server, cron and lpsched) w shows me: "stuff about uptime" 1.00 1.00 1.00 Is ist normal to have a load average of at least 1.00 ? At my company we have a SUN 3/50 it displays almost 0 (around 0.10) when nothing happens. Also the Public Domain load average prg that has reached me just several days ago display the same values as /usr/bin/w. Could anyone please answer that ? And if it's NOT normal to have at least 1.00 what is wrong ? Thanks, Axel -- Domain: fischer@netmbx.UUCP Europe: ...!tmpmbx!netmbx!fischer Rest of world: ...!uunet!pyramid!tmpmbx!netmbx!fischer =====> Beam me up, Scotty - there is no intelligent life down here ! <=====
davidsen@sungod.crd.ge.com (William Davidsen) (06/13/89)
In article <2910@netmbx.UUCP> fischer@netmbx.UUCP (Axel Fischer) writes: | Is ist normal to have a load average of at least 1.00 ? At my company we have | a SUN 3/50 it displays almost 0 (around 0.10) when nothing happens. Not on either of my systems... on a quiet system I do see all zeros from time to time. Sounds as though you have a hungry "idle daemon" running. Do you by any chance leave vpix running? bill davidsen (davidsen@crdos1.crd.GE.COM) {uunet | philabs}!crdgw1!crdos1!davidsen "Stupidity, like virtue, is its own reward" -me
etienne@accsys.UUCP (Stefan Hauser) (06/14/89)
In article <745@crdgw1.crd.ge.com> davidsen@crdos1.UUCP (bill davidsen) writes: >In article <2910@netmbx.UUCP> fischer@netmbx.UUCP (Axel Fischer) writes: > >| Is ist normal to have a load average of at least 1.00 ? At my company we have >| a SUN 3/50 it displays almost 0 (around 0.10) when nothing happens. > > Not on either of my systems... on a quiet system I do see all zeros >from time to time. Sounds as though you have a hungry "idle daemon" >running. Do you by any chance leave vpix running? on my 386'er with SCO xenix 2.3.1, after booting. all shows 0 if i do nothing, but after heavy load, it goes only back to 1 and next time there are a lot of jobs, it stays by 2... seem like a little bug...or am i wrong? (there are always the same processes running, if no one is on the system... but the load average wont go back to 0...) etienne (sorry for my bad english...but i never learned it...i do it...perhaps next year...) -- ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Stefan Hauser | CH-Sargans | SHORT : etienne@accsys.UUCP BANG : ....!mcvax!cernvax!impch!accsys!etienne OR : ....!uunet!acad!acadch!impch!accsys!etienne (<- faster from usa)
fischer@netmbx.UUCP (Axel Fischer) (06/15/89)
In article <411@accsys.UUCP+ etienne@accsys.UUCP (Stefan Hauser) writes: +In article <745@crdgw1.crd.ge.com> davidsen@crdos1.UUCP (bill davidsen) writes: +>In article <2910@netmbx.UUCP> fischer@netmbx.UUCP (Axel Fischer) writes: +>> Is ist normal to have a load average of at least 1.00 ? At my company we have +>> a SUN 3/50 it displays almost 0 (around 0.10) when nothing happens. +> Not on either of my systems... on a quiet system I do see all zeros +>from time to time. Sounds as though you have a hungry "idle daemon" +>running. Do you by any chance leave vpix running? +on my 386'er with SCO xenix 2.3.1, after booting. all shows 0 if i do nothing, +but after heavy load, it goes only back to 1 and next time there are a lot +of jobs, it stays by 2... seem like a little bug...or am i wrong? +(there are always the same processes running, if no one is on the system... +but the load average wont go back to 0...) That seems the right answer. I have just installed OS Release 2.3.2 and it all started again. First a load average of 0 than after heavy load a load average of 1 and so on. There is surely a bug in w. Maybe SCO reads this and fixed it to the next release or it is already fixed in 2.3.3. SCO: Thanks for fixing the select() call bug from 2.3.1 to 2.3.2 ! Now all works correct - maybe me *countless* mails to rosso@sco has helped. -Axel -- Domain: fischer@netmbx.UUCP Europe: ...!tmpmbx!netmbx!fischer Rest of world: ...!uunet!pyramid!tmpmbx!netmbx!fischer =====> Beam me up, Scotty - there is no intelligent life down here ! <=====
nick@qtnet.UUCP (Nick Lawes) (06/22/89)
>That seems the right answer. I have just installed OS Release 2.3.2 and it >all started again. First a load average of 0 than after heavy load a load >average of 1 and so on. > >There is surely a bug in w. >Maybe SCO reads this and fixed it to the next release or it is already fixed >in 2.3.3. Hmm.. I must have skipped the earlier discussions, but suddenly got interested. I am running SCO Xenix V 386 r 2.3.2, and thought I'd see if I got the same problem. Just thought you may be interested in a report... In short, I don't. I use 'w' and a little program that I hacked up to get _avenrun from the kernel, and they both reset to zero after high load averages (if you call 7 high enough). >SCO: Thanks for fixing the select() call bug from 2.3.1 to 2.3.2 ! Now >all works correct - maybe me *countless* mails to rosso@sco has >helped. That's interesting... Having read in the installation guide (I think) that the select() call was now supported, I thought I'd investigate.. The first thing I noticed was a total lack of mention of it in the documentation (or am I missing something?). I finally found the call by scanning the libraries. Slibtinfo.a seems a strange place for a system call interface, but still, ours not to reason why... Having never used select() before, I turned to the SUN 3 and with the aid of the manuals wrote a test program to poll a couple of fifo's. This worked fine! Okay, move the code to the Xenix box and try to run it there... No way will it work! I get no errno set after the select() call, so maybe the select() call in tinfo probably the right one after all. But where is it? I've searched all the libraries to no avail... I'd be interested in learning where I'm going wrong... Can anyone out there enlighten me? Point me at the right manual / page? Thanks in advance for any help... -- Nick -- [ Nick Lawes, Systems Programmer | voice: +44 1 353 6723 ] [ Technical Marketing, Quotnet (UK) Ltd. | email: nick@quotnet.co.uk ] [ 12 Norwich Street, London. EC4a 1BP | email: ..!mcvax!ukc!qtnet!nick ] [ | ham : G8ZHR @ GB7UWS ]
csch@netcs.UUCP (Clemens Schrimpe) (07/08/89)
nick@qtnet.UUCP (Nick Lawes) writes:
<> That's interesting... Having read in the installation guide (I think)
<> that the select() call was now supported, I thought I'd investigate..
<> The first thing I noticed was a total lack of mention of it in the
<> documentation (or am I missing something?). I finally found the call
<> by scanning the libraries. Slibtinfo.a seems a strange place for a
<> system call interface, but still, ours not to reason why...
Hmmm ... since support for select() came up first in 2.3.1 it's
naturally mentioned first in the 2.3.1 Development System, which is currently
shipping as I've heard.
<> Having never used select() before, I turned to the SUN 3 and with the
<> aid of the manuals wrote a test program to poll a couple of fifo's.
<> This worked fine! Okay, move the code to the Xenix box and try to run
<> it there... No way will it work!
I't won't even work on SunOS 4.x (I tried) !
This is because fifo's in BSD are sockets, while they are 'pipes' in SYS-V!
The select always shows, that the fifo is 'readable' on SYS-V (and SunOS 4.x) !
Clemens Schrimpe
--
UUCP: csch@netcs BITNET: csch@db0tui6.BITNET
ARPA/NSF: csch@garp.mit.edu PSI: PSI%45300033047::CSCH
PHONE: +49-30-24 42 37 FAX: +49-30-24 38 00
BTX: 0303325016-0003 TELEX: 186672 net d