Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 27 Mar 2009 14:18:02 +0100
From:      Adrian Penisoara <ady@freebsd.ady.ro>
To:        Pierre Lamy <pierre@userid.org>
Cc:        freebsd-net@freebsd.org, Shawn Everett <shawn@tandac.com>
Subject:   Re: FreeBSD Router Problem
Message-ID:  <78cb3d3f0903270618u229986c1idddd0e5548444c60@mail.gmail.com>
In-Reply-To: <49CBA72F.3020600@userid.org>
References:  <3650.206.108.16.89.1235691792.squirrel@alder.hosix.com> <3853.206.108.16.89.1235693214.squirrel@alder.hosix.com> <78cb3d3f0902261619t71a054fet43779c37e2981603@mail.gmail.com> <200902262341.35069.shawn@tandac.com> <49CAB28A.9030406@userid.org> <1865.206.108.16.89.1238019698.squirrel@alder.hosix.com> <78cb3d3f0903260552g372fd4b6k886bba1ebc05a77c@mail.gmail.com> <49CBA72F.3020600@userid.org>

next in thread | previous in thread | raw e-mail | index | archive | help
Hi,

On Thu, Mar 26, 2009 at 5:02 PM, Pierre Lamy <pierre@userid.org> wrote:

> states        hard limit    10000
>
> If I want to dos this box all I need to do is hold 10k tcp connections open
> in established.
>
> A 1 day default timeout for established connections is retarded, since
> virtually all client apps and OSs as well as intervening stateful firewalls
> will lose state after 1 hour. A session which is idle for more than an hour
> can't be considered to be active. Coupled with an extremely low state limit,
> and you're asking for problems. If the session is active at all before the
> session timeout is hit, the timer is reset.


I'm sorry but I have to object. Having past experience in Oracle Support for
networking issues I did see many problems with statefull firewalls which
were cutting off idle Oracle connections. The base line is: DO NOT assume
connections are dead even if they are idle for more than an hour...


>
>
> I'm not saying he's getting DOSd, but with such low limits, even a normal
> home network is going to run into problems at some point. We can see from
> the diagnostic output provided earlier that there were no issues when it was
> collected, but was it collected while there was an outage?



>
> If the problem still occurs, it may be worth scripting something to collect
> some pfctl -g -v -v -v -s all and some sysctl -a, vmstat output as well.


Well, just keep a 'pfctl -s state >/var/tmp/pf-states.txt' running in cron
every few minutes then and let's check it out...

Regards,
Adrian.


>
>
> Pierre
>
> Adrian Penisoara wrote:
>
>> Hi,
>>
>> On Wed, Mar 25, 2009 at 11:21 PM, Shawn Everett <shawn@tandac.com> wrote:
>>
>>
>>
>>> tcp.established           86400s
>>>>
>>>> ^^ This should be 3600.
>>>>
>>>> Pierre
>>>>
>>>>
>>> That's an interesting thought.  Why would that matter?
>>>
>>>
>>
>>
>> It's the PF TCP established session timeout, which defaults to 1 day. This
>> is relevant only if you see a lot of ESTABLISHED sessions in the 'pfctl -s
>> state' output, which appears not to be the case...
>>
>>
>> Regards,
>> Adrian.
>> _______________________________________________
>> freebsd-net@freebsd.org mailing list
>> http://lists.freebsd.org/mailman/listinfo/freebsd-net
>> To unsubscribe, send any mail to "freebsd-net-unsubscribe@freebsd.org"
>>
>>
>



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?78cb3d3f0903270618u229986c1idddd0e5548444c60>