Return to BSD News archive
Path: euryale.cc.adfa.oz.au!newshost.anu.edu.au!harbinger.cc.monash.edu.au!news.mel.connect.com.au!munnari.OZ.AU!news.ecn.uoknor.edu!paladin.american.edu!gatech!newsfeed.internetmci.com!howland.reston.ans.net!math.ohio-state.edu!magnus.acs.ohio-state.edu!lerc.nasa.gov!purdue!haven.umd.edu!news.umbc.edu!cs.umd.edu!not-for-mail From: torek@elf.bsdi.com (Chris Torek) Newsgroups: comp.unix.admin,comp.unix.bsd.bsdi.misc Subject: Re: mystery page faults Date: 19 Jan 1996 18:23:40 -0800 Organization: Berkeley Software Design, Inc. Lines: 75 Message-ID: <4dpjnc$al5@elf.bsdi.com> References: <4dmmpt$g0k@sleipnir.iaccess.com.au> Reply-To: torek@bsdi.com NNTP-Posting-Host: elf.bsdi.com Xref: euryale.cc.adfa.oz.au comp.unix.admin:37578 comp.unix.bsd.bsdi.misc:2213 In article <4dmmpt$g0k@sleipnir.iaccess.com.au> <phil@iaccess.com.au> writes: >I have these mysterious page faults showing up. None of the faults here seem mysterious to me... >bash# vmstat -w 5 > procs memory page disks faults cpu > r b w avm fre flt re pi po fr sr wd0 wd1 in sy cs us sy id > 0 0 0 0 12964 112 13 0 0 0 0 3 0 158 124 15 4 3 93 This shows 112 page faults, 13 of them being `reclaims'. A bit of explanation from vmstat(1): page Information about page faults and paging activity. These are av- eraged each five seconds, and given in units per second. re page reclaims (simulating reference bits) This is not quite right. On the VAX, reclaims were used to simulate reference bits; now, reclaims are simply reclaims, i.e., we took access to the page away because you had not used it (or had released it, e.g., the binary exited), and then you used it again anyway (e.g., ran the binary again) while it was still valid. at pages attached (found in free list) The number under `re' is more or less this statistic (i.e., the man page is wrong). pi pages paged in po pages paged out fr pages freed per second de anticipated short term memory shortfall sr pages scanned by clock algorithm, per-second The `de' parameter does not exist in the current VM system, which has no swap-in (nor swap-out) code. Some of the other statistics (well, really just `fr') are also not recorded since the VM is really still mostly just the old Mach 2.5 VM. Under `faults', we have: faults Trap/interrupt rate averages per second over last 5 seconds. in device interrupts per interval (including clock inter- rupts) sy system calls per interval cs cpu context switch rate (switches/interval) Hence this shows 158 device interrupts, 124 system calls (open, stat, read, etc.), and 15 context switches. The system is mostly idle: cpu Breakdown of percentage usage of CPU time. us user time for normal and low priority processes sy system time id cpu idle A total of 7% of the available CPU cycles were spent on useful work and/or overhead, 4% in user mode and 3% in system (kernel) mode; the remaining 93% of the cycles were spent in the idle loop. The only other number that shows up is here: > procs memory page disks faults cpu > r b w avm fre flt re pi po fr sr wd0 wd1 in sy cs us sy id > 0 0 0 0 13684 117 6 4 0 0 0 5 0 191 440 17 1 4 95 This time we had four pages paged in, probably when a process ran and referred to its pages for the first time. In no case did the available free pages (memory `fre') fall below about 12 megabytes. This means there was never any need to push anything to swap space. -- In-Real-Life: Chris Torek, Berkeley Software Design Inc El Cerrito, CA Domain: torek@bsdi.com +1 510 234 3167