Interesting analysis of linux kernel threading by IBM

[Jason Gunthorpe]

On Thu, 20 Jan 2000, Larry McVoy wrote:

> This is a completely unsubstatiated statement.  OK, everyone, do this:
> 
> $ vmstat
> load free cach swap pgin  pgou  dk0 dk1 dk2 dk3 ipkt opkt  int  ctx  usr sys idl
> 0.00  4.3 96.9  0.4    0     0    0   0   0   0    0    0  104   53    0   3  97
> 
> Those people with the load field higher than 10, please tell us what you are
> doing.

Hum, its not quite 10 right now (and hopefully won't be again soon) It did
hit 40 the other day mind you:

saens{jgg}~#vmstat   
   procs                      memory    swap          io     system        cpu
 r  b  w   swpd   free   buff  cache  si  so    bi    bo   in    cs  us sy id
 4  3  0   3396   1628  14816 127560   0   0     1     1    0     2   1 1  2
saens{jgg}~#ps aux | wc -l
    182
saens{jgg}~#w
  7:00pm  up 187 days,  2:36,  1 user,  load average: 6.34, 4.39, 3.29
saens{jgg}~#procinfo
Linux 2.2.10-ac10 (root en saens) (gcc 2.7.2.3) #2 1CPU [saens.(none)]

Memory:      Total        Used        Free      Shared     Buffers Cached
Mem:        258164      255652        2512      198252        5904 117388
Swap:       118448        3396      115052

Bootup: Sat Jul 17 17:23:25 1999    Load average: 6.21 5.16 3.73 1/188
22237

user  :  22d 11:50:33.70  12.0%  page in :3070841179  disk 1: 937666119r291450156w
nice  :       3:33:00.93   0.1%  page out:491514704
system:  16d  7:50:14.17   8.7%  swap in :   568142
idle  : 148d  3:25:23.25  79.2%  swap out:    32739
uptime: 187d  2:39:11.89         context :2308759798

This machine is ftp.debian.org - two days ago it sent out a more than 180G
of traffic (145G via FTP, rest HTTP/rsync) over a 1 day period. Which is
what, 1/7th the size of cdrom.com? We run Apache and wu-ftpd (though we
switched to proftpd two days ago..) Machine is PII-350, 256Meg ram and a
3x9G RAID5 on a ICP Vortex RAID controller 

The biggest load generator is really rsync, followed by proftpd, both when
they do ls -LR's :< I don't get a feeling that schedualing is a big
concern here.

Enjoy,
Jason


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo en vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/