Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 6 Nov 2009 11:01:24 -0500
From:      "Kevin" <k@kevinkevin.com>
To:        "'no name'" <britneyfreek@googlemail.com>
Cc:        freebsd-pf@freebsd.org
Subject:   RE: Question about round robin
Message-ID:  <00a501ca5efa$65640890$302c19b0$@com>
In-Reply-To: <-3431979369893017739@unknownmsgid>
References:  <00a201ca5ef6$7a4f3ee0$6eedbca0$@com> <-3431979369893017739@unknownmsgid>

Next in thread | Previous in thread | Raw E-Mail | Index | Archive | Help


> -----Original Message-----
> From: no name [mailto:britneyfreek@googlemail.com]
> Sent: Friday, November 06, 2009 10:52 AM
> To: Kevin
> Cc: freebsd-pf@freebsd.org
> Subject: Re: Question about round robin
>=20
> i am not quite sure but i assume pf does no availabililtty checks as
> responses from systems behind an address might take too long...
>=20
> - regards
>=20
> Am 06.11.2009 um 16:34 schrieb "Kevin" <k@kevinkevin.com>:
>=20
> > Hello,
> >
> >
> >
> > I am curious about simple round robin load balancing w/ PF. I see
> > how simple
> > it is to setup within the FreeBSD configuration file, but have a
> > fairly
> > simple question.
> >
> > Does PF detect if one of the addresses in the "pool" is not
> > responsive? Or
> > does it just blindly send traffic to all the addresses in the pool
> > regardless? Is there some sort of heartbeat functionality or status
> > check
> > perhaps? Has anyone accomplished something along these lines? Load
> > balancing
> > would be pretty useless if I couldn't take a server out of the pool
> > without
> > having to re-configure PF every time.
> >
> >
> > Any comments are appreciated!
> >
> >
> > Thanks,
> >
> >
> > Kevin
> > www.stardothosting.com


I've searched the freebsd-* mailing list as well as gone through the =
handbook regarding load balancing w/ PF =
(http://www.openbsd.org/faq/pf/pools.html) and found no mention of any =
kind of inherent availability checks or anything along those lines. Have =
I missed something?=20

Seems that all the options (bitmask, random, source-hash, round-robin) =
provide for different ways to distribute traffic to the servers, but if =
a server dies or becomes unresponsive it would compromise the pool in =
itself. Most other load balancing solutions such as LVS can be =
incorporated with keepalived to allow for status checking.

I would love to know anyone who may have implemented a solution like =
that with PF + round robin.


Thanks,

Kevin




Want to link to this message? Use this URL: <http://docs.FreeBSD.org/cgi/mid.cgi?00a501ca5efa$65640890$302c19b0$>