On 4/20/06, Florin Malita <[EMAIL PROTECTED]> wrote:
> De obicei exista o corelatie directa intre frecventa intreruperilor HW
> si procesarea softirq, asa ca un cat /proc/interrupts ar trebui sa-ti
> dea o idee.
Pai nu zice mare lucru (cel putin pt mine):
<pre>
[EMAIL PROTECTED] ~]# cat /proc/interrupts
CPU0 CPU1
0: 216223 339991 IO-APIC-edge timer
8: 0 0 IO-APIC-edge rtc
9: 0 0 IO-APIC-level acpi
16: 31006 5714695 IO-APIC-level eth0
18: 9376 98577 IO-APIC-level libata, eth2
19: 8304034 26854 IO-APIC-level skge, eth1
NMI: 357 336
LOC: 556192 556169
ERR: 1
MIS: 0
[EMAIL PROTECTED] ~]#
</pre>
> > PS: am eu o alta problema mai mare pe care am urmarit'o pana in
> > punctul asta: intreruperi soft. Daca vreti, dau si detalii :)
> >
> Please.
Masina este pe post de firewall.
MB: Asus p5wd2-premium
CPU: P4 630D la 3.0GHz, cu HT, 64 biti)
OS: FC5
eth0: intel gigabit on board
eth1: intel gigabit pe pci
eth2: intel 100 pe pci
eth3: marvel gigabit on board (driver:skge) --> aici banui eu problema...
Trafic maxim ajunge seara la ~150-200mbps.
la mpstat:
Average: CPU %user %nice %sys %iowait %irq %soft
%idle intr/s
Average: all 0.30 0.00 0.50 0.20 1.00 4.90
93.11 6874.85
Average: 0 0.40 0.00 1.00 0.20 1.00 4.99
92.42 4149.30
Average: 1 0.20 0.00 0.20 0.00 1.00 4.79
93.61 2725.35
(vad o corespondenta intre nr de pachete pe secunda si nr de
intreruperi pe secunda.... banui ca un pachet genereaza o intrerupere.
spun prostii?)
Da, am irqbalance activat si HT activat din bios.
[EMAIL PROTECTED] htb]# ps aux | grep irq
root 3 0.0 0.0 0 0 ? SN 17:26 0:00 [ksoftirqd/0]
root 6 0.0 0.0 0 0 ? SN 17:26 0:00 [ksoftirqd/1]
root 1718 0.0 0.0 2708 300 ? Ss 17:26 0:00 irqbalance
root 9765 0.0 0.0 56812 708 pts/2 S+ 18:22 0:00 grep irq
[EMAIL PROTECTED] htb]# uname -a
Linux cerberus 2.6.16-1.2096_FC5 #1 SMP Wed Apr 19 05:14:26 EDT 2006
x86_64 x86_64 x86_64 GNU/Linux
[EMAIL PROTECTED] htb]# vmstat 1 10
procs -----------memory---------- ---swap-- -----io---- --system-- ----cpu----
r b swpd free buff cache si so bi bo in cs us sy id wa
1 0 0 1642328 39120 249084 0 0 36 32 3401 95 0 6 93 1
0 0 0 1642204 39128 249076 0 0 0 128 9059 180 0 10 89 0
0 0 0 1642204 39132 249072 0 0 0 0 8254 96 0 7 93 0
0 0 0 1642204 39140 249064 0 0 0 340 7297 70 0 5 95 0
0 0 0 1642204 39140 249064 0 0 0 0 7170 66 0 8 92 0
0 0 0 1641956 39164 249040 0 0 0 72 8405 128 0 5 94 0
0 0 0 1641956 39168 249036 0 0 4 0 8069 151 0 6 93 1
0 0 0 1642328 39168 249036 0 0 0 0 8509 48 0 7 93 0
0 0 0 1642452 39168 249036 0 0 0 0 8881 42 0 9 91 0
0 0 0 1642452 39168 249036 0 0 0 0 8472 42 0 9 92 0
[EMAIL PROTECTED] htb]#
Problema e ca uneori ingheata complet. Eu banui ca e de la trafic prea
mare pe placa Marvell integrata.
Ce am incercat: sa opresc tot pe el si sa'l las pe post de sarma: am
oprit firewall, qos, squid, named, zebra, bgpd, si merge fara greturi.
Cand le pornesc la loc, dupa un timp aleator moare. Nici ping nu merge
in el. Dupa reset de la buton totul e ok. In loguri nimic. Parca ar fi
din triunghiul bermudelor, nimic nu s'o intamplat.
Sugestii? Sa'l arunc pe geam?
PS: vechea masina era un celeron la 2.4, mb: asus p4s800-se si mergea
de n-avea aer, dar uneori ajungea incarcarea procesorului la 30-45% si
de asta l-am "upgradat".
--
...and justice for all...
_______________________________________________
RLUG mailing list
[email protected]
http://lists.lug.ro/mailman/listinfo/rlug
