The kclick process is the proprietary routing daemon that Perfigo/CCA uses. It captures all available resources on the CAS. When we first got clean machines, the load averages on the machines made me nausious :) my gernealy rule for other *NIX boxen being load = # of CPU's (in a perfect world)....
Are you noticing a significant performance hit for subnets protected by CCA?
Ryan Dorman, CCNP
Netowrk Communications Specialist
Communications and Network Services
Millerville University
-----Original Message-----
From: Perfigo SecureSmart and CleanMachines Discussion List on behalf of Michael Grinnell
Sent: Thu 9/1/2005 4:49 PM
To: [log in to unmask]
Subject: Re: [PERFIGO] Https 404's and High Load Average
We have the same model, but with 2GBs of RAM running in virtual
gateway mode. Our CASs seem to be running ok. Your load avg does
seem a bit high, but I've noticed on the CAS that it always seems to
run at 100% busy. Here's our numbers:
4:36pm up 12 days, 5:08, 1 user, load average: 2.55, 2.30, 2.22
93 processes: 90 sleeping, 3 running, 0 zombie, 0 stopped
CPU states: 1.1% user, 98.8% system, 0.0% nice, 0.0% idle
Mem: 2059480K av, 728240K used, 1331240K free, 60K shrd,
50744K buff
Swap: 1048552K av, 0K used, 1048552K free
404620K cached
PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
1167 root 14 0 0 0 0 RW 92.7 0.0 17243m kclick
30325 root 14 0 1012 1012 752 R 3.3 0.0 0:01 top
1475 root 9 0 193M 193M 14900 S 0.9 9.6 94:55 java
22982 root 9 0 193M 193M 14900 S 0.1 9.6 0:00 java
1 root 8 0 520 520 452 S 0.0 0.0 0:04 init
and from the event log heartbeat
Sep 1 16:12:22 <snip> Perfigo: CleanAccessServer:192.168.10.10
System Stats: Load factor 0 (max since reboot: 477) Mem
(bytes) Total: 2108907520 Used: 743907328 Free: 1365000192 Shared:
61440 Buffers: 51953664 Cached: 413302784 CPU User: 1% Nice: 0%
System: 99% Idle: 0%
Sep 1 16:12:22 <snip> Perfigo: CleanAccessServer:192.168.10.20
System Stats: Load factor 1 (max since reboot: 325) Mem
(bytes) Total: 2108907520 Used: 568860672 Free: 1540046848 Shared:
61440 Buffers: 51179520 Cached: 294789120 CPU User: 0% Nice: 0%
System: 100% Idle: 0%
I would look at other factors than just load. You might take a look
at the apache logs (/perfigo/access/apache/logs/) and the tomcat logs
(/perfigo/access/tomcat/logs) for more clues.
Also, Gigabit Ethernet does not support manual duplex and speed
negotiation, IIRC. You are supposed to use auto. Here's our
modules.conf
alias eth0 bcm5700
alias eth1 bcm5700
alias parport_lowlevel parport_pc
<snip>
Hope that helps,
Michael Grinnell
Network Security Administrator
The American University
e-mail: [log in to unmask]
On Sep 1, 2005, at 4:18 PM, Brad Kramer wrote:
> Sorry, should have mentioned that I have broadcoms, in the
> modules.conf, we
> hardset the speed/duplex 1000 and full....
>
> I honestly think this has something to do with load....
>
> Here's the first 20 or so lines of top
>
> 4:15pm up 20 days, 23:31, 1 user, load average: 3.40, 3.49, 3.47
> 99 processes: 96 sleeping, 3 running, 0 zombie, 0 stopped
> CPU states: 2.6% user, 97.3% system, 0.0% nice, 0.0% idle
> Mem: 3597396K av, 728352K used, 2869044K free, 56K shrd,
> 55248K
> buff
> Swap: 1048552K av, 0K used, 1048552K free
> 485632K
> cached
>
> PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
> 1183 root 14 0 0 0 0 RW 90.1 0.0 29273m kclick
> 18026 root 14 0 1016 1016 752 R 3.0 0.0 0:03 top
> 1479 root 9 0 106M 106M 14892 S 1.9 3.0 65:35 java
> 16 root 9 0 0 0 0 SW 0.7 0.0 0:51
> kjournald
> 566 root 9 0 616 616 508 S 0.3 0.0 0:58 syslogd
> 1 root 8 0 520 520 452 S 0.0 0.0 0:08 init
> 2 root 8 0 0 0 0 SW 0.0 0.0 0:00 keventd
> 3 root 19 19 0 0 0 RWN 0.0 0.0 0:00
> ksoftirqd_CPU0
> 4 root 9 0 0 0 0 SW 0.0 0.0 0:00 kswapd
> 5 root 9 0 0 0 0 SW 0.0 0.0 0:00
> kreclaimd
> 6 root 9 0 0 0 0 SW 0.0 0.0 0:00 bdflush
> 7 root 9 0 0 0 0 SW 0.0 0.0 0:00 kupdated
> 8 root -1 -20 0 0 0 SW< 0.0 0.0 0:00
> mdrecoveryd
> 131 root 9 0 0 0 0 SW 0.0 0.0 0:00
> kjournald
> 571 root 9 0 1108 1108 444 S 0.0 0.0 0:00 klogd
> 713 root 9 0 1252 1252 1112 S 0.0 0.0 0:00 sshd
> 765 root 9 0 1272 1272 904 S 0.0 0.0 0:00 nessusd
> 783 root 8 0 676 676 568 S 0.0 0.0 0:00 crond
> 819 daemon 9 0 572 572 488 S 0.0 0.0 0:00 atd
> 1060 root 9 0 936 936 860 S 0.0 0.0 0:00
> _plutorun
> 1061 root 9 0 400 400 340 S 0.0 0.0 0:00 logger
> 1065 root 9 0 936 936 860 S 0.0 0.0 0:00
> _plutorun
>
> A little farther down we have about 20 instances of java...
>
> Let me know what you think...
> Thanks
> -Brad Kramer
>
>
> On 9/1/05 4:02 PM, "Simon Bell" <[log in to unmask]> wrote:
>
>
>> check your speed and duplex? ifconfig on the CAS/CAM
>>
>>
>>>>> [log in to unmask] 9/1/2005 10:17 AM >>>
>>>>>
>> Here at Ashland we also had a "successful" rollout of CCA this
>> fall... We
>> are currently running 3.5.4, and we have experienced a small hiccup.
>>
>> Most users are getting the initial redirect page just fine,
>> however the
>> secure web server times out on many requests, and when it does
>> respond it is
>> extremely slow....
>>
>> Currently we have roughly 1000 users behind an inline CAS.
>> The CAS is running on a Compaq DL360 3.4ghz with 4G RAM....
>> I think it is due to high load on the server, because the load
>> average sits
>> around 2.5 all day (through slow times and busy times) our RAM is
>> virtually
>> untouched...
>> I have contacts at other universities that are running 5x more
>> users behind
>> a single CAS with no problems...... Any suggestions?
>>
>> Thanks!
>>
>> -------------
>> Bradley W. Kramer
>> Network/Telecom Intern.
>> Ashland University
>> (419) 289-5630
>> [log in to unmask]
>>
>
> -------------
> Bradley W. Kramer
> Network/Telecom Intern.
> Ashland University
> (419) 289-5630
> [log in to unmask]
>
|
