vmstat on ulab, 11/13/03, for CS444

u18(1)% uptime

12:39pm up 12 day(s), 5:50, 2 users, load average: 1.03, 0.81, 0.43

time period for load averageà 1 min 5 min 15 min

12 days = 12*24*60*60 = 1.04M secs

u18(2)% vmstat -s

0 swap ins

0 swap outs

0 pages swapped in

0 pages swapped out

16216571 total address trans. faults taken

276606 page ins almost all by PFs (following demand paging)

268417 page outs

445982 pages paged in looks like usually 2 pages/PF

272747 pages paged out

3573934 total reclaims

3573667 reclaims from free list

0 micro (hat) faults

16216571 minor (as) faults page in physical mem, but PTE not set up

271651 major faults page not in phys. mem, “real” PF

4073128 copy-on-write faults

2630980 zero fill page faults

0 pages examined by the clock daemon

0 revolutions of the clock hand

272410 pages freed by the clock daemon

71817 forks

53679 vforks

134482 execs

82335553 cpu context switches

457395550 device interrupts 457M ints/1Msec = 457 ints/sec

27064490 traps

2332097941 system calls 233M syscalls/1Msecs = 233 syscalls/sec

57615826 total name lookups (cache hits 92%)

30011258 user cpu

50347672 system cpu

25198560 idle cpu

221871 wait cpu

Here we see 445, 982 pages paged in, or about 3 per exec. Lots of memory sharing for shell, etc.

Similar today, also 0 revolutions of clock hand

System users.cs.umb.edu = blade71,  today, Nov. 21, 2006: a fairly busy system

blade71(1)% uptime

 11:44am  up 177 day(s), 16:29,  13 users,  load average: 0.04, 0.04, 0.06

(177 days = 4248 hours = 15.3 x 10^6 secs)

blade71(2)% vmstat -s
    27090 swap ins
    24637 swap outs
    54180 pages swapped in
 32520802 pages swapped out
198524678 total address trans. faults taken
 67337825 page ins
 27645975 page outs
 83407679 pages paged in
 41671226 pages paged out
 36022278 total reclaims
 35979070 reclaims from free list
        0 micro (hat) faults
198524678 minor (as) faults
 66978270 major faults
 28009962 copy-on-write faults
 28920404 zero fill page faults
1364690101 pages examined by the clock daemon
    90096 revolutions of the clock hand            sounds big, but maybe 900/day for 100 days, about 1/hour
 68884840 pages freed by the clock daemon
   937661 forks
    35712 vforks
   609255 execs
2475158889 cpu context switches
2477787244 device interrupts
311764576 traps
2341821352 system calls
2693108977 total name lookups (cache hits 96%)
 14295818 user   cpu
 12164200 system cpu
1454539831 idle   cpu
 54209127 wait   cpu

vmstat -10 for 10-second reports, after first line.

r = in run queue

b = blocked

w = swapped

re = reclaims from free list

mf = major faults

pi = page ins

po = page outs

fr, de, …not so important

in = non-clock interrupts

sy = syscalls

cs = context switches, i.e. process switches

us = % user

sy = % kernel

id = % idle

u18(9)% vmstat 10

procs memory page disk faults cpu

r b w swap free re mf pi po fr de sr dd f0 s1 -- in sy cs us sy id

0 0 0 865040 382144 3 15 3 2 2 0 0 3 0 0 0 332 2226 77 28 48 24

0 0 0 865080 377384 0 0 0 0 0 0 0 0 0 0 0 306 12 38 0 0 100

0 0 0 865080 377360 0 0 0 3 3 0 0 1 0 0 0 334 66 52 0 0 100

0 0 0 865000 377288 0 0 0 0 0 0 0 3 0 0 0 317 17 42 0 1 99

0 0 0 865080 377360 0 0 0 0 0 0 0 1 0 0 0 322 54 49 0 0 100

0 0 0 865352 377616 0 9 0 0 0 0 0 1 0 0 0 331 53475 60 6 12 82

0 0 0 865224 377464 1 8 1 0 0 0 0 3 0 0 0 316 303780 47 36 64 0

0 0 0 865240 377464 0 0 0 0 0 0 0 0 0 0 0 323 306180 57 36 64 0

0 0 0 865240 377456 0 0 0 0 0 0 0 0 0 0 0 326 306147 54 35 65 0

0 0 0 865240 377440 0 0 0 0 0 0 0 1 0 0 0 308 304049 44 38 62 0

0 0 0 865048 377240 0 0 0 0 0 0 0 0 0 0 0 334 305403 83 35 65 0

0 0 0 865240 377432 0 0 0 2 2 0 0 1 0 0 0 321 306289 45 36 64 0

0 0 0 865240 377456 0 0 0 4 4 0 0 1 0 0 0 313 304492 42 38 62 0

0 0 0 865240 377464 0 0 0 0 0 0 0 0 0 0 0 323 306061 55 37 63 0

Suspect NSF server activity for busy period—backup?