Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 24 Mar 2008 16:41:37 -0300
From:      AT Matik <asstec@matik.com.br>
To:        "Alexander Shulikov" <shulikov@gmail.com>
Cc:        freebsd-ipfw@freebsd.org, bu7cher@yandex.ru
Subject:   Re: kern/121955: [ipfw] [panic] freebsd 7.0 panic with mpd
Message-ID:  <200803241641.37884.asstec@matik.com.br>
In-Reply-To: <18292fe60803241208i248ffeachf12c49aee3ab2756@mail.gmail.com>
References:  <18292fe60803240107v1462a87v4222790745844d5d@mail.gmail.com> <200803241411.51930.asstec@matik.com.br> <18292fe60803241208i248ffeachf12c49aee3ab2756@mail.gmail.com>

next in thread | previous in thread | raw e-mail | index | archive | help
On Monday 24 March 2008 16:08:10 Alexander Shulikov wrote:
> By default I have:
> # sysctl kern.ipc.nmbclusters
> kern.ipc.nmbclusters: 25600

hard to say, do you checked netstat -m if you get to your limit? If you get=
=20
there set it higher

> # sysctl net.inet.ip.intr_queue_maxlen
> net.inet.ip.intr_queue_maxlen: 50

seems to be the default
try 128, 256, 512

>
> What range of value is optimal to try?
> Also I add to loader.conf:
> kern.maxusers=3D1536

I guess maxusers does not help for your setup and probably you should let t=
he=20
system selftune itself


> kern.ipc.maxpipekva=3D32000000
> net.graph.maxalloc=3D2048
> (but it was added after panic's)
>

I believe both do not help anything for ipfw


> Other thing, that I want to try - net.isr.direct -> 0? May be it
> temprorary resolved problem, because packet will be going to queue for
> processing.
>

I guess this does not help dummynet either




> 2008/3/24, AT Matik <asstec@matik.com.br>:
> > On Monday 24 March 2008 12:11:44 Alexander Shulikov wrote:
> >  > In real script I have in and out. But some ip's for the same speed I
> >  > add in table, and then do:
> >  > # 512/128
> >  > ${fwcmd} pipe 357 config bw 128Kbit/s queue 100 mask src-ip 0xffffff=
ff
> >  > ${fwcmd} pipe 358 config bw 512Kbit/s queue 100 mask dst-ip 0xffffff=
ff
> >  > ${fwcmd} add pipe 357 ip from "table(3)" to any in
> >  > ${fwcmd} add pipe 358 ip from any to "table(3)" out
> >  >
> >  > IPs for individual speed added as in example in previous letter, but
> >  > with in/out.
> >  >
> >  > But what can I do for resolve this problem. Now I have server with
> >  > FreeBSD 6.2 and copy of this configs - and all works fine. o_O
> >
> > well I use the above different, first I define the pipe and second the =
bw
> >  anyway I do not use tables
> >
> >  also seems you use long queues so may be you like to tune
> >  net.inet.ip.intr_queue_maxlen and mbuf of your machine
> >
> >  > 2008/3/24, AT Matik <asstec@matik.com.br>:
> >  > > On Monday 24 March 2008 11:51:44 Alexander Shulikov wrote:
> >  > >  > # sysctl -a | grep one_pass
> >  > >  > net.inet.ip.fw.one_pass: 0
> >  > >  >
> >  > >  > Yes - it eq 0. But I need it for next situation: all net I need
> >  > >  > shape at one speed, but invididual ip addresses to another spee=
d.
> >  > >  > For example,
> >  > >  > ipfw pipe 1 config bw 10Mbit/s queue 100
> >  > >  > ipfw pipe 2 config bw 10Mbit/s queue 100
> >  > >  > ipfw add pipe 1 ip from 192.168.1.0/24 to any
> >  > >  > ipfw add pipe 2 ip from any to 192.168.1.0/24
> >  > >  > ipfw pipe 3 config bw 1Mbit/s queue 100
> >  > >  > ipfw pipe 4 config bw 1Mbit/s queue 100
> >  > >  > ipfw add pipe 3 ip from 192.168.1.1/32 to any
> >  > >  > ipfw add pipe 4 ip from any to 192.168.1.1/32
> >  > >
> >  > > that should work, I have similar setups running fine
> >  > >  the /32 mask you should not need but I am missing the in/out
> >  > > definition in your rules
> >  > >
> >  > >  > ......
> >  > >  >
> >  > >  >
> >  > >  > Also this configuration work in FreeBSD 6.2. (May be in 6.2
> >  > >  > smaller call tree?)
> >  > >  >
> >  > >  > 2008/3/24, AT Matik <asstec@matik.com.br>:
> >  > >  > > On Monday 24 March 2008 08:08:02 Andrey V. Elsukov wrote:
> >  > >  > >  > AT Matik wrote:
> >  > >  > >  > > what do you mean? By setting to 0 the packages are not
> >  > >  > >  > > re-injected into the pipe but go through other existing
> >  > >  > >  > > rules after the matching pipe, or not?
> >  > >  > >  >
> >  > >  > >  > When you reset net.inet.ip.fw.one_pass to zero, packets
> >  > >  > >  > return back into ipfw to the next rule after
> >  > >  > >  > dummynet/netgraph. And if you have similar rules packets
> >  > >  > >  > will be passed into
> >  > >  > >  > dummynet/netgraph again.
> >  > >  > >  >
> >  > >  > >  > This is example how to get double fault (from mail archive=
):
> >  > >  > >
> >  > >  > > jaaa well but that is the famous bw 0 example which is not
> >  > >  > > valid, as by itself certainly an invalid config, not connected
> >  > >  > > to the existing problem the reporter has I guess
> >  > >  > >
> >  > >  > >  Jo=C3=A3o
> >  > >  > >
> >  > >  > >  > ifconfig em0 192.168.0.2/24
> >  > >  > >  > kldload ipfw
> >  > >  > >  > kldload dummynet
> >  > >  > >  > sysctl net.inet.ip.fw.one_pass=3D0
> >  > >  > >  > ipfw pipe 2 config bw 0
> >  > >  > >  > ipfw add 2 pipe 2 ip from any to any
> >  > >  > >  > ipfw add 2 pipe 2 ip from any to any
> >  > >  > >  > ipfw add 2 pipe 2 ip from any to any
> >  > >  > >  > ipfw add 2 pipe 2 ip from any to any
> >  > >  > >  > ipfw add 2 pipe 2 ip from any to any
> >  > >  > >  > ipfw add 2 pipe 2 ip from any to any
> >  > >  > >  > ipfw add 2 pipe 2 ip from any to any
> >  > >  > >  > ipfw add 2 pipe 2 ip from any to any
> >  > >  > >  > ipfw add 2 pipe 2 ip from any to any
> >  > >  > >  > ipfw add 2 pipe 2 ip from any to any
> >  > >  > >  > ipfw add 2 pipe 2 ip from any to any
> >  > >  > >  > ipfw add 2 pipe 2 ip from any to any
> >  > >  > >  > ipfw add 2 pipe 2 ip from any to any
> >  > >  > >  > ipfw add 2 pipe 2 ip from any to any
> >  > >  > >  > ipfw add 2 pipe 2 ip from any to any
> >  > >  > >  > ipfw add 2 pipe 2 ip from any to any
> >  > >  > >  > ipfw add 2 pipe 2 ip from any to any
> >  > >  > >  > ping 192.168.0.1
> >  > >  > >
> >  > >  > >  --
> >  > >  > >
> >  > >  > >
> >  > >  > > Atenciosamente, J.M.
> >  > >  > >  Respons=C3=A1vel Plant=C3=A3o Site Support Matik
> >  > >  > >  Infomatik Internet Technology
> >  > >  > >  (18)3551.8155  (18)8112.7007
> >  > >  > >  http://info.matik.com.br
> >  > >  > >
> >  > >  > >
> >  > >  > >
> >  > >  > >
> >  > >  > >
> >  > >  > >
> >  > >  > >
> >  > >  > >  A mensagem foi scaneada pelo sistema de e-mail e pode ser
> >  > >  > > considerada segura. Service fornecido pelo Datacenter Matik
> >  > >  > > https://datacenter.matik.com.br
> >  > >
> >  > > --
> >  > >
> >  > >
> >  > >  Atenciosamente, J.M.
> >  > >  Respons=C3=A1vel Plant=C3=A3o Site Support Matik
> >  > >  Infomatik Internet Technology
> >  > >  (18)3551.8155  (18)8112.7007
> >  > >  http://info.matik.com.br
> >  > >
> >  > >
> >  > >
> >  > >
> >  > >
> >  > >
> >  > >
> >  > >  A mensagem foi scaneada pelo sistema de e-mail e pode ser
> >  > > considerada segura. Service fornecido pelo Datacenter Matik
> >  > > https://datacenter.matik.com.br
> >
> > --
> >
> >
> >  Atenciosamente, J.M.
> >  Respons=C3=A1vel Plant=C3=A3o Site Support Matik
> >  Infomatik Internet Technology
> >  (18)3551.8155  (18)8112.7007
> >  http://info.matik.com.br
> >
> >
> >
> >
> >
> >
> >
> >  A mensagem foi scaneada pelo sistema de e-mail e pode ser considerada
> > segura. Service fornecido pelo Datacenter Matik=20
> > https://datacenter.matik.com.br

=2D-=20

Atenciosamente, J.M.
Respons=C3=A1vel Plant=C3=A3o Site Support Matik
Infomatik Internet Technology
(18)3551.8155 =C2=A0(18)8112.7007
http://info.matik.com.br







A mensagem foi scaneada pelo sistema de e-mail e pode ser considerada segura.
Service fornecido pelo Datacenter Matik  https://datacenter.matik.com.br



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200803241641.37884.asstec>