top output broked?
PID USERNAME PRI NICE SIZERES STATE C TIME WCPUCPU COMMAND 824 root -80 1048K 596K biord 0 0:38 0.00% 0.00% find 385 root 40 32740K 31944K select 1 0:32 0.00% 0.00% XFree86 836 root -80 532K 276K biord 1 0:07 0.00% 0.00% nfsd 14848 root 960 26912K 26832K RUN1 0:04 0.00% 0.00% ld 424 bright 40 2120K 1340K select 0 0:04 0.00% 0.00% rxvt no cpu time, known issue? -- -Alfred Perlstein - [[EMAIL PROTECTED]|[EMAIL PROTECTED]] Represent yourself, show up at BABUG http://www.babug.org/ To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: top output broked?
Date: Tue, 27 Mar 2001 03:18:10 -0800 From: Alfred Perlstein [EMAIL PROTECTED] PID USERNAME PRI NICE SIZERES STATE C TIME WCPUCPU COMMAND 824 root -80 1048K 596K biord 0 0:38 0.00% 0.00% find 385 root 40 32740K 31944K select 1 0:32 0.00% 0.00% XFree86 836 root -80 532K 276K biord 1 0:07 0.00% 0.00% nfsd 14848 root 960 26912K 26832K RUN1 0:04 0.00% 0.00% ld 424 bright 40 2120K 1340K select 0 0:04 0.00% 0.00% rxvt no cpu time, known issue? I get non-zero values from time to time; in particular, I fired up an xterm did a "while (1)" loop in it, and the CPU times increased in a gratifying manner. :-} However, the usual values I'm seeing are rather lower than I would expect, and lower than the same machine running -STABLE (within the last several days, by my recollection). As a reality check, I'm trying "vmstat 5", and it's consistently reporting either 99 or 100% idle. There -- I got both it top to report something noticeable: I fired up netscape Maybe it really *is* using CPU much more efficiently...? No, I didn't think so, but it was a nice thought :-) Oh: recent CVSup history (I hadn't noticed the behavior in the -CURRENNT I built yesterday): CVSup started at Sun Mar 25 23:47:00 PST 2001 CVSup ended at Sun Mar 25 23:52:25 PST 2001 CVSup started at Mon Mar 26 23:47:00 PST 2001 CVSup ended at Mon Mar 26 23:53:39 PST 2001 Cheers, david -- David H. Wolfskill [EMAIL PROTECTED] As a computing professional, I believe it would be unethical for me to advise, recommend, or support the use (save possibly for personal amusement) of any product that is or depends on any Microsoft product. To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
RE: top output broked?
On 27-Mar-01 Alfred Perlstein wrote: PID USERNAME PRI NICE SIZERES STATE C TIME WCPUCPU COMMAND 824 root -80 1048K 596K biord 0 0:38 0.00% 0.00% find 385 root 40 32740K 31944K select 1 0:32 0.00% 0.00% XFree86 836 root -80 532K 276K biord 1 0:07 0.00% 0.00% nfsd 14848 root 960 26912K 26832K RUN1 0:04 0.00% 0.00% ld 424 bright 40 2120K 1340K select 0 0:04 0.00% 0.00% rxvt no cpu time, known issue? Not one that I've seen: PID USERNAME PRI NICE SIZERES STATE C TIME WCPUCPU COMMAND 11 root -160 0K 0K CPU0 0 79.5H 49.37% 49.37% idle: cpu0 10 root -160 0K 0K RUN1 79.4H 48.19% 48.19% idle: cpu1 13 root -48 -167 0K 0K WAIT 0 62:53 0.00% 0.00% swi6: tty:s 15 root 760 0K 0K sleep 0 6:07 0.00% 0.00% random 5 root 200 0K 0K syncer 1 2:47 0.00% 0.00% syncer 20 root -68 -187 0K 0K WAIT 1 1:18 0.00% 0.00% irq18: fxp0 19 root -64 -183 0K 0K WAIT 0 0:53 0.00% 0.00% irq16: ahc0 12 root -44 -163 0K 0K WAIT 0 0:52 0.00% 0.00% swi1: net 18 root -36 -155 0K 0K WAIT 1 0:49 0.00% 0.00% swi3: cambi 4 root -160 0K 0K psleep 0 0:41 0.00% 0.00% bufdaemon 283 root 40 552K 388K select 0 0:10 0.00% 0.00% dhclient If you run 'top -S' does all your time show up in the idle processes like it does here? -- John Baldwin [EMAIL PROTECTED] -- http://www.FreeBSD.org/~jhb/ PGP Key: http://www.baldwin.cx/~john/pgpkey.asc "Power Users Use the Power to Serve!" - http://www.FreeBSD.org/ To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
RE: top output broked?
Date: Tue, 27 Mar 2001 08:33:10 -0800 (PST)
From: John Baldwin [EMAIL PROTECTED]
Not one that I've seen:
PID USERNAME PRI NICE SIZERES STATE C TIME WCPUCPU COMMAND
11 root -160 0K 0K CPU0 0 79.5H 49.37% 49.37% idle: cpu0
10 root -160 0K 0K RUN1 79.4H 48.19% 48.19% idle: cpu1
13 root -48 -167 0K 0K WAIT 0 62:53 0.00% 0.00% swi6: tty:s
15 root 760 0K 0K sleep 0 6:07 0.00% 0.00% random
5 root 200 0K 0K syncer 1 2:47 0.00% 0.00% syncer
20 root -68 -187 0K 0K WAIT 1 1:18 0.00% 0.00% irq18: fxp0
19 root -64 -183 0K 0K WAIT 0 0:53 0.00% 0.00% irq16: ahc0
12 root -44 -163 0K 0K WAIT 0 0:52 0.00% 0.00% swi1: net
18 root -36 -155 0K 0K WAIT 1 0:49 0.00% 0.00% swi3: cambi
4 root -160 0K 0K psleep 0 0:41 0.00% 0.00% bufdaemon
283 root 40 552K 388K select 0 0:10 0.00% 0.00% dhclient
If you run 'top -S' does all your time show up in the idle processes like it
does here?
Hmm... mine loks like that (modulo #CPUs), except when I'm actually
making it do some work (re-building the kernel, in this case). What I
see ("top -S") looks like:
last pid: 9546; load averages: 0.97, 0.64, 0.30up 0+00:08:32 08:51:47
77 processes: 3 running, 57 sleeping, 2 zombie, 15 waiting
CPU states: 91.1% user, 0.0% nice, 5.4% system, 0.4% interrupt, 3.1% idle
Mem: 32M Active, 78M Inact, 27M Wired, 96K Cache, 35M Buf, 110M Free
Swap: 1024M Total, 4168K Used, 1020M Free
PID USERNAME PRI NICE SIZERES STATETIME WCPUCPU COMMAND
10 root -160 0K 0K RUN 3:43 2.15% 2.15% idle
9545 root 1210 3428K 3304K RUN 0:00 2.00% 0.10% cc1
514 root 40 27148K 26128K select 0:03 0.00% 0.00% XFree86
212 root 40 420K 304K select 0:03 0.00% 0.00% pccardd
610 david 40 4200K 3332K select 0:01 0.00% 0.00% xterm
12 root -48 -167 0K 0K WAIT 0:01 0.00% 0.00% swi6: tty:sio
7601 root 80 5112K 4764K wait 0:01 0.00% 0.00% make
16 root -64 -183 0K 0K WAIT 0:01 0.00% 0.00% irq14: ata0
312 root 40 976K 556K select 0:00 0.00% 0.00% moused
603 david 40 2348K 1968K select 0:00 0.00% 0.00% ssh
11 root -44 -163 0K 0K WAIT 0:00 0.00% 0.00% swi1: net
1193 david 960 1964K 1208K RUN 0:00 0.00% 0.00% top
5 root 200 0K 0K syncer 0:00 0.00% 0.00% syncer
288 root 40 2256K 1504K select 0:00 0.00% 0.00% sshd
620 david 40 4200K 3332K select 0:00 0.00% 0.00% xterm
14 root 760 0K 0K sleep0:00 0.00% 0.00% random
350 root -68 -187 0K 0K WAIT 0:00 0.00% 0.00% irq3: an0
I confess a degree of skepticism :-}
Here's output from "vmstat -5" around that time:
1 2 0 43184104456 306 0 0 0 121 0 0 0 402 599 419 99 1 0
1 2 0 37772109324 1209 0 0 0 1498 0 1 0 432 2032 549 91 6 2
1 2 0 43032104124 918 0 0 0 669 0 1 0 394 1053 422 96 4 0
1 2 0 44540102584 289 0 0 0 188 0 0 0 348 402 307 99 1 0
2 1 0 39272108032 1123 0 0 0 1446 0 8 0 356 1515 362 90 5 5
1 2 0 38448107844 1047 0 0 0 1066 0 10 0 368 1428 388 88 5 7
1 1 0 36596108380 1240 0 0 0 1310 0 1 0 354 1643 362 94 5 2
2 1 0 32264110776 1334 0 0 0 1504 0 2 0 381 1851 431 92 6 2
1 0 0 30376110236 1850 0 0 0 1892 0 7 0 389 2506 462 86 8 6
1 2 0 34904107064 1786 0 0 0 1692 0 4 0 360 2339 394 88 7 5
2 0 0 30040108884 2437 0 0 0 2634 0 12 0 384 3194 466 77 10 14
1 2 0 35544104636 1885 0 0 0 1734 0 27 0 405 2556 490 81 9 10
1 2 0 34528104520 2339 0 0 0 2432 0 3 0 378 3136 455 85 10 5
1 1 0 34620103844 2066 0 0 0 2115 0 2 0 370 2787 432 91 8 1
procs memory pagedisks faults cpu
r b w avm fre flt re pi po fr sr ad0 md10 in sy cs us sy id
2 0 0 28532106216 2618 0 0 0 2876 0 8 0 394 3524 514 83 13 4
2 0 0 36172102488 2071 0 0 0 1967 0 7 0 402 2910 507 89 9 2
Cheers,
david
--
David H. Wolfskill [EMAIL PROTECTED]
As a computing professional, I believe it would be unethical for me to
advise, recommend, or support the use (save possibly for personal
amusement) of any product that is or depends on any Microsoft product.
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message
Re: top output broked?
Date: Tue, 27 Mar 2001 10:21:46 -0800 (PST)
From: John Baldwin [EMAIL PROTECTED]
Hmm... mine loks like that (modulo #CPUs), except when I'm actually
making it do some work (re-building the kernel, in this case). What I
see ("top -S") looks like:
last pid: 9546; load averages: 0.97, 0.64, 0.30up 0+00:08:32
08:51:47
77 processes: 3 running, 57 sleeping, 2 zombie, 15 waiting
CPU states: 91.1% user, 0.0% nice, 5.4% system, 0.4% interrupt, 3.1% idle
This is probably right..
Yes; that much of it "feels" about right.
I don't know why you are seeing such weirdness however. Is your world and
kernel out of sync.
Assuredly not, but I understand the rationale behind the question. :-)
(I have the "script" log available for perusal)
It's a nice (mis)feature now that if items in the middle
of kinfo_proc change size it still tries to use the misordered data rather than
complaining about it like it used to. :-P See my other e-mail where top on my
laptop doles out time to userland tasks ok.
I confess a degree of skepticism :-}
I agree.
:-)
Date: Tue, 27 Mar 2001 10:21:45 -0800 (PST)
From: John Baldwin [EMAIL PROTECTED]
Keep in mind that we no longer charge interrupt time to the process being
interrupted, instead all that interrupt handling has been pushed off into
ithreads. Same for software interrupt threads.
OK; that's a good useful thing to keep in mind. And I did see some
IRQ-related entries in top's output.
That said, I don't see how X is so idle, it's certainly not on my laptop:
PID USERNAME PRI NICE SIZERES STATETIME WCPUCPU COMMAND
454 john 40 0K 43464K select 1:57 4.05% 4.05% XFree86
461 john 40 17076K 16144K select 0:35 0.39% 0.39% enlightenment
492 john 4 10 3072K 2040K select 0:28 0.10% 0.10% E-ScreenSave.
Eh... the "enlightenment" line may provide a clue there. I use tvtwm as
a window manager. :-} (I figure anything that could be marginally
acceptable on a (maxed out) 24 MB Sun 3/60 ought to be adequate for this
750 MHz/256 MB laptop)
Cheers,
david
--
David H. Wolfskill [EMAIL PROTECTED]
As a computing professional, I believe it would be unethical for me to
advise, recommend, or support the use (save possibly for personal
amusement) of any product that is or depends on any Microsoft product.
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message
Re: top output broked?
Also, I happened to note that as I'm doing a "make buildworld" (for today's -STABLE, running in yesterday's -STABLE), my "top -S" output shows a large number of "0.00" entries for CPU (on the same laptop as my previously-reported results). So it may be odd, but at least -- in my case -- it appears to be moderately consistent (modulo known expected changes between -STABLE -CURRENT). Cheers, david -- David H. Wolfskill [EMAIL PROTECTED] As a computing professional, I believe it would be unethical for me to advise, recommend, or support the use (save possibly for personal amusement) of any product that is or depends on any Microsoft product. To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: top output broked?
On 27-Mar-01 David Wolfskill wrote: Date: Tue, 27 Mar 2001 10:21:45 -0800 (PST) From: John Baldwin [EMAIL PROTECTED] Keep in mind that we no longer charge interrupt time to the process being interrupted, instead all that interrupt handling has been pushed off into ithreads. Same for software interrupt threads. OK; that's a good useful thing to keep in mind. And I did see some IRQ-related entries in top's output. Are they getting %CPU though. When running top -S, the CPU %'s should always add up to about 100 (with fudges for rounding errors). That said, I don't see how X is so idle, it's certainly not on my laptop: PID USERNAME PRI NICE SIZERES STATETIME WCPUCPU COMMAND 454 john 40 0K 43464K select 1:57 4.05% 4.05% XFree86 461 john 40 17076K 16144K select 0:35 0.39% 0.39% enlightenment 492 john 4 10 3072K 2040K select 0:28 0.10% 0.10% E-ScreenSave. Eh... the "enlightenment" line may provide a clue there. I use tvtwm as a window manager. :-} (I figure anything that could be marginally acceptable on a (maxed out) 24 MB Sun 3/60 ought to be adequate for this 750 MHz/256 MB laptop) Heh, but I figured Alfred was in X when he was running top, so X must've been doing _some_ screen updates, and not just have 0.00% CPU time. :-P -- John Baldwin [EMAIL PROTECTED] -- http://www.FreeBSD.org/~jhb/ PGP Key: http://www.baldwin.cx/~john/pgpkey.asc "Power Users Use the Power to Serve!" - http://www.FreeBSD.org/ To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: top output broked?
Date: Tue, 27 Mar 2001 11:56:38 -0800 (PST) From: John Baldwin [EMAIL PROTECTED] OK; that's a good useful thing to keep in mind. And I did see some IRQ-related entries in top's output. Are they getting %CPU though. When running top -S, the CPU %'s should always add up to about 100 (with fudges for rounding errors). Well, as noted in another note a little prior to this one, the -CURRENT behavior I'm seeing isn't all *that* different from the -STABLE behavior -- in each case, the sum of what "top" reports for CPU % is normally small. Eh... the "enlightenment" line may provide a clue there. I use tvtwm as a window manager. :-} (I figure anything that could be marginally acceptable on a (maxed out) 24 MB Sun 3/60 ought to be adequate for this 750 MHz/256 MB laptop) Heh, but I figured Alfred was in X when he was running top, so X must've been doing _some_ screen updates, and not just have 0.00% CPU time. :-P Well, that gets into a matter of perspective, since the amount of CPU resource required to do the screen updates (vs. what is available) could well be 0.00 (to 2 decimals) :-) (Kinda like the ratio of a circle's circumference to its diameter is "3" to a single significant figure.) (I was in X at the time, too.) Cheers, david -- David H. Wolfskill [EMAIL PROTECTED] As a computing professional, I believe it would be unethical for me to advise, recommend, or support the use (save possibly for personal amusement) of any product that is or depends on any Microsoft product. To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: top output broked?
On 27-Mar-01 David Wolfskill wrote: Date: Tue, 27 Mar 2001 11:56:38 -0800 (PST) From: John Baldwin [EMAIL PROTECTED] OK; that's a good useful thing to keep in mind. And I did see some IRQ-related entries in top's output. Are they getting %CPU though. When running top -S, the CPU %'s should always add up to about 100 (with fudges for rounding errors). Well, as noted in another note a little prior to this one, the -CURRENT behavior I'm seeing isn't all *that* different from the -STABLE behavior -- in each case, the sum of what "top" reports for CPU % is normally small. -STABLE doesn't have idle processes. :) Eh... the "enlightenment" line may provide a clue there. I use tvtwm as a window manager. :-} (I figure anything that could be marginally acceptable on a (maxed out) 24 MB Sun 3/60 ought to be adequate for this 750 MHz/256 MB laptop) Heh, but I figured Alfred was in X when he was running top, so X must've been doing _some_ screen updates, and not just have 0.00% CPU time. :-P Well, that gets into a matter of perspective, since the amount of CPU resource required to do the screen updates (vs. what is available) could well be 0.00 (to 2 decimals) :-) (Kinda like the ratio of a circle's circumference to its diameter is "3" to a single significant figure.) (I was in X at the time, too.) Fair enough.. -- John Baldwin [EMAIL PROTECTED] -- http://www.FreeBSD.org/~jhb/ PGP Key: http://www.baldwin.cx/~john/pgpkey.asc "Power Users Use the Power to Serve!" - http://www.FreeBSD.org/ To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: top output broked?
On Tue, Mar 27, 2001 at 03:18:10AM -0800, Alfred Perlstein wrote: PID USERNAME PRI NICE SIZERES STATE C TIME WCPUCPU COMMAND 824 root -80 1048K 596K biord 0 0:38 0.00% 0.00% find 385 root 40 32740K 31944K select 1 0:32 0.00% 0.00% XFree86 836 root -80 532K 276K biord 1 0:07 0.00% 0.00% nfsd 14848 root 960 26912K 26832K RUN1 0:04 0.00% 0.00% ld 424 bright 40 2120K 1340K select 0 0:04 0.00% 0.00% rxvt Hmm, I just rebuilt world recently (this morning), and I'm seeing this: PID USERNAME PRI NICE SIZERES STATE C TIME WCPUCPU COMMAND 311 root 1260 600K 264K CPU1 0 24:11 52.39% 52.39% nfsd 503 postfix40 1604K 904K select 0 0:02 0.15% 0.15% qmgr 8069 root 960 2088K 1240K CPU0 0 0:00 0.00% 0.00% top Eek. nfsd is sucking up CPU, even while it's idle (its only nfs client is down for a few quick repairs). I think it's been doing this since the TI-RPC stuff was imported. But, back to your problem, no 0.0% CPU time problem here. - alex To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
