server load: your opinion?

[nimbius]

so i run the following server as a mailserver at work. postfix, saslauthd, MailScanner, mysql, httpd, dovecot.
spamassassin, clamav, php

HP DL145 G2
dual 64 bit 1ghz processors

processor : 0
vendor_id : AuthenticAMD
cpu family : 15
model : 37
model name : AMD Opteron(tm) Processor 252
stepping : 1
cpu MHz : 1000.000
cache size : 1024 KB
fpu : yes
fpu_exception : yes
cpuid level : 1
wp : yes
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 syscall nx mmxext fxsr_opt lm 3dnowext 3dnow pni lahf_lm
bogomips : 2008.99
TLB size : 1024 4K pages
clflush size : 64
cache_alignment : 64
address sizes : 40 bits physical, 48 bits virtual
power management: ts fid vid ttp

processor : 1
vendor_id : AuthenticAMD
cpu family : 15
model : 37
model name : AMD Opteron(tm) Processor 252
stepping : 1
cpu MHz : 1000.000
cache size : 1024 KB
fpu : yes
fpu_exception : yes
cpuid level : 1
wp : yes
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 syscall nx mmxext fxsr_opt lm 3dnowext 3dnow pni lahf_lm
bogomips : 2008.99
TLB size : 1024 4K pages
clflush size : 64
cache_alignment : 64
address sizes : 40 bits physical, 48 bits virtual
power management: ts fid vid ttp

4 GB memory
dual GBnics teamed


free memory available is usually aroung 1-2 gigs.

we process 5000 messages per day (including spam connect attempts that are RBL'd)
during peak use i see:

12:43:39 up 2 days, 12 min, 3 users, load average: 6.35, 3.99, 2.61

these load avg's are high i think. high enough to slow authentication against the server and make VI take a good minute or two to return to console after ":wq"

normally i see something closer to
13:11:31 up 2 days, 40 min, 3 users, load average: 3.11, 2.25, 1.92


whats your opinion? shouldnt i see something lower than this?

[BillT440]

this is no scientific solution by any means, but I've noticed keeping the run queue at no higher than a count of 4 per processor core keeps the systems very responsive to load spikes . This is just based on information gathered and experiences here at work. Your results will vary. I guess it all depends on what your processes are doing. Waiting on I/O or CPU?

[pete_1967]

You're trying to run awfully amny services from one box. Spamassassin is high on resource usage especially if you use remote checks as well (check their site for optimisation options) and MySQL uses all available resources unless you configure it to restrictive settings.

Combine 2 of those with Apache desperately trying to serve (I assume) dynamic content with Perl/ PhP or which ever it is you are using and clamav fighting for resources so it can scan the files.

Load averages, imho, are not that high but keep eye on I/O (disk access) and all that BillT440 said.

[nimbius]

then i would have to agree, IO access to the two 500GB SATA drives is a culprit. load averages spiked to 8 and 9 today, i assume while sending a 130mb queue to a user who just never reported his client as nonfunctional. still, this spike is surprising given as its just one client receiving (again, mysql, httpd servers and a local DNS can account for the spike occuring as well)

This is a mailserver by trade, and the initial assertion regarding spamassassin was correct. ive removed spamassassin and clamav in exchange for RBL checks and virus content scanning by the client. Users are grumpy about having a whitelist now, but the RBL budget on the system is far more forgiving than running spamd with comprable if not better performance.

i guess im just really surprised to see a dual processor, 64 bit system grind to a halt this easily...then again perhaps my expectations are a bit high. 64 bit computing has been touted, imho too heavily the past few years. lets not even get into how many cores the big 2 want us to have.