Thanks Ray and thanks Brad, too!
At 09:28 07.06.2002 -0700, you wrote:
>Possible? Yeah sure, I suppose it is "possible". But you'd do better to
>give us a more systematic profile of what the router is doing if you want
>good opinions.
OK, the router is used as a firewall and dhcpd and tinydns and dnscache in
front of a small net of five computers, where only two are used as
web-frontends. There is a squid running (but not used yet) and a sshd.
I hope the webcam is not the reason, because there are requests to it like
max. 20 per minute. And this is the only traffic, but the load average
shows
# w
13:56:40 up 0 Days (17h), load average: 0.81 0.79 0.79
>The "load average" numbers that various apps report are sort of odd
>things. They don't represent a true system "load", at least not to my
>thinking; instead, they represent the average number of processes that are
>waiting for some resource to be allocated to them (that is, blocked
>processes). Unless you are running some unusual userspace firewalling app
>on your system, I wouldn't expect basic routing demands to increase system
>load by this measure.
>
>So ... in your guess, substitute "blocking" for "looping" and you are
>probably right. But what process? Hmmm ... here are some things to
>consider (and perhaps tell us about):
>
>1. How does CPU utilization change with this change in load? If you have
>"top" available, it calculates and reports the numbers I have in mind; if
>not, you'll need to get the raw data from /proc/stat and do the arithmetic
>yourself.
First, I have to correct: The Machine is equiped with 48MBytes RAM, not 64....
OK, here's the top-output:
# top
3:45pm up 1:02, 1 user, load average: 1.82, 1.68, 1.39
36 processes: 35 sleeping, 1 running, 0 zombie, 0 stopped
CPU states: 24.6% user, 25.7% system, 0.0% nice, 50.2% idle
Mem: 46828K av, 33608K used, 13220K free, 15720K shrd, 14228K buff
Swap: 0K av, 0K used, 0K free 6032K cached
PID USER PRI NI SIZE RSS SHARE STAT LIB %CPU %MEM TIME COMMAND
2643 root 19 0 820 820 660 R 0 31.8 1.7 11:12 top
1963 root 0 0 1148 1148 988 S 0 7.7 2.4 2:40 sshd
1285 root 4 0 444 444 384 S 0 3.9 0.9 0:24 pppoe
3725 root 0 0 324 324 268 S 0 2.8 0.6 0:00 sleep
1862 root 0 0 904 904 744 S 0 1.6 1.9 0:31 ipmail
1854 squid 1 0 4148 4148 1160 S 0 1.2 8.8 0:20 squid
1851 root 0 0 1288 1288 1100 S 0 0.5 2.7 0:10 xntpd
1771 root 0 0 356 356 296 S 0 0.2 0.7 0:05 svscan
1 root 0 0 468 468 408 S 0 0.1 0.9 0:08 init
2 root 0 0 0 0 0 SW 0 0.0 0.0 0:00 kflushd
3 root 0 0 0 0 0 SW 0 0.0 0.0 0:00 kupdate
4 root 0 0 0 0 0 SW 0 0.0 0.0 0:00 kswapd
5 root 0 0 0 0 0 SW 0 0.0 0.0 0:00 keventd
6 root -20 -20 0 0 0 SW< 0 0.0 0.0 0:00 mdrecoveryd
1276 root 0 0 960 960 768 S 0 0.0 2.0 0:00
adsl-connect
1284 root 0 0 792 792 664 S 0 0.0 1.6 0:03 pppd
1504 root 0 0 552 552 460 S 0 0.0 1.1 0:01 syslogd
1513 root 0 0 496 496 372 S 0 0.0 1.0 0:00 klogd
1520 root 0 0 484 484 420 S 0 0.0 1.0 0:00 inetd
1527 root 0 0 864 864 760 S 0 0.0 1.8 0:01 sshd
1532 root 0 0 272 272 224 S 0 0.0 0.5 0:00 watchdog
1537 root 0 0 572 572 484 S 0 0.0 1.2 0:00 cron
1596 root 0 0 952 952 808 S 0 0.0 2.0 0:00 squid
1601 root 0 0 728 728 584 S 0 0.0 1.5 0:00 dhcpd
1833 root 0 0 308 308 260 S 0 0.0 0.6 0:00 supervise
1834 root 0 0 308 308 260 S 0 0.0 0.6 0:00 supervise
1835 tinydns 0 0 332 332 268 S 0 0.0 0.7 0:00 tinydns
1836 dnslog 0 0 308 308 256 S 0 0.0 0.6 0:00 multilog
1857 root 0 0 308 308 260 S 0 0.0 0.6 0:00 supervise
1858 root 0 0 308 308 260 S 0 0.0 0.6 0:00 supervise
1859 dnscache 0 0 1588 1588 340 S 0 0.0 3.3 0:04 dnscache
1860 dnslog 0 0 364 364 312 S 0 0.0 0.7 0:01 multilog
1863 root 0 0 460 460 392 S 0 0.0 0.9 0:00 getty
1864 root 0 0 460 460 392 S 0 0.0 0.9 0:00 getty
1871 squid 0 0 296 296 248 S 0 0.0 0.6 0:00 unlinkd
1970 root 0 0 1184 1184 952 S 0 0.0 2.5 0:00 sh
during that, the w-command gives me:
# w
15:45:12 up 0 Days (1h), load average: 2.17 1.76 1.41
USER TTY PID TIMEON FROM
root ttyp1 1970 58 nl-ws-boris.nordlabs.network
Later I had the following output, where the pppoe is quite busy....
I also saw the ipmail increase CPU-usage towards 10%.
6:05pm up 3:22, 1 user, load average: 2.69, 2.78, 2.50
40 processes: 33 sleeping, 7 running, 0 zombie, 0 stopped
CPU states: 28.5% user, 46.3% system, 0.0% nice, 25.8% idle
Mem: 46828K av, 33856K used, 12972K free, 16456K shrd, 14228K buff
Swap: 0K av, 0K used, 0K free 6036K cached
PID USER PRI NI SIZE RSS SHARE STAT LIB %CPU %MEM TIME COMMAND
1285 root 20 0 444 444 384 R 0 38.6 0.9 29:20 pppoe
6308 root 16 0 820 820 660 R 0 22.0 1.7 12:36 top
6254 root 0 0 1148 1148 984 S 0 5.4 2.4 3:18 sshd
>2. What userspace apps are you normally running on the system?
I hope you can see that in the top-output, too.... Otherwise: What do you
mean with 'userspace apps'?
>3. Does "the dns is no longer reliable" refer to a DNS server on the
>router, or do you just mean that the slowness of the line causes offsite
>DNS requests to time out (or do you mean something different from either
>of these)?
On the (Win2k-) Client I got time-outs resoving the hostnames!
>4. What apps does "top" report as high-usage?
see 1.
>5. What filesystem hardware does the router use? (Blocking can easily be
>due to the need to access a slow hard disk, for example.) In particular,
>might syslogd be blocking somehow (it is a natural one to think of as
>creeping up over time)?
It's a RAMdisk.
>All of this isn't even up to the level of fishing yet; it is just some
>preliminary thoughts about a problem on a vaguely described router and network.
OK, but I'm quite thankful even for that, for I have no more ideas....
I will - as long as my ability allows - do everything to bring a light into
this.
Thanks again!
Boris
>At 05:07 PM 6/7/02 +0200, Boris Andratzek wrote:
>>Hej All!
>>
>>Since several Weeks I got the following problem, that doesn't seem to be
>>normal.
>>I use a dachstein 1.02.1 with glibc 2.1.3 on CD and the hardware is a IBM
>>PC 330 (P-166) with 64 MBytes RAM ant two Realtek 8139 NICs. The system
>>is stable and doing everything I want, but:
>>After the (re)start the load average is as I am used to from some other
>>machines:
>># w
>> 16:53:06 up 0 Days (0h), load average: 0.04 0.01 0.02
>>
>>After a non-predictable period of time the load increases as if one
>>process begins looping or so:
>># w
>> 17:14:39 up 0 Days (0h), load average: 0.79 0.86 0.81
>>
>>and even later:
>># w
>> 17:31:12 up 0 Days (0h), load average: 2.02 1.11 0.81
>>
>>In this state the dns is no longer reliable and the bandwith is lower
>>than it could be.
>>
>>So, what do you think?? What can I try to do??
>>I yet changed the hardware once to a second PC of the same type and
>>actually to an other type of IBM PC (P-75). Everything stays the same. Is
>>it possible that this high load comes from requesting my apache behind
>>the firewall quite frequently (WebCam)?
>
>
>--
>-----------------------------------------------"Never tell me the
>odds!"--------------
>Ray Olszewski -- Han Solo
>Palo Alto, California, USA [EMAIL PROTECTED]
>-------------------------------------------------------------------------------------------
>
>
>_______________________________________________________________
>
>Don't miss the 2002 Sprint PCS Application Developer's Conference
>August 25-28 in Las Vegas -- http://devcon.sprintpcs.com/adp/index.cfm
>
>------------------------------------------------------------------------
>leaf-user mailing list: [EMAIL PROTECTED]
>https://lists.sourceforge.net/lists/listinfo/leaf-user
>SR FAQ: http://leaf-project.org/pub/doc/docmanager/docid_1891.html
_______________________________________________________________
Don't miss the 2002 Sprint PCS Application Developer's Conference
August 25-28 in Las Vegas -
http://devcon.sprintpcs.com/adp/index.cfm?source=osdntextlink
------------------------------------------------------------------------
leaf-user mailing list: [EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/leaf-user
SR FAQ: http://leaf-project.org/pub/doc/docmanager/docid_1891.html
