Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 28 Nov 2003 08:41:04 -0200
From:      "Daniel C. Sobral" <dcs@tcoip.com.br>
To:        Bruce M Simpson <bms@spc.org>
Cc:        cvs-all@FreeBSD.org
Subject:   Re: cvs commit: src/etc rc.network src/etc/defaults rc.conf src/share/man/man5 rc.conf.5
Message-ID:  <3FC72640.9060305@tcoip.com.br>
In-Reply-To: <20031127213110.GA76702@saboteur.dek.spc.org>
References:  <200311270951.hAR9pT19090236@repoman.freebsd.org> <3FC65A71.3050104@tcoip.com.br> <20031127213110.GA76702@saboteur.dek.spc.org>

next in thread | previous in thread | raw e-mail | index | archive | help
Bruce M Simpson wrote:

>>What if /usr/local is NFS mounted?
>=20
> Up to the user where they put their isakmpd or racoon package. It is, a=
fter
> all, relocatable for such a reason. This merely brings in the infrastru=
cture
> to make running it possible at the right time.
>=20
> This would certainly be the case in an embedded [wireless] system.

I'm sorry, but I think an NFS mounted /usr is far more common than an=20
NFS directory mounted over IPSEC. I advance that this commit priviledges =

an unusual setup over a more common one. We do install isakmpd on=20
/usr/local by default, after all. If a directory depends on isakmpd=20
being up, it shouldn't be auto-mounted through fstab, IMHO.

But, alas, what rcNG *does not* do, which is it's greatest flaw IMO, is=20
taking into account network dependencies correctly. Network dependencies =

change too much to have it statically ordered, beyond a certain point.

One example is the case above. There are reasonable grounds for wanting=20
isakmpd to be up both before and after NFS. Before if you want to mount=20
NFS through IPSEC, and after if you keep isakmpd NFS-mounted, and=20
doesn't care for encrypted NFS mounts.

Another example is dynamic routers and ntpd. OSPF is very=20
time-sensitive, and clock changes can play hell with the routing tables, =

so ntpd ought to be up before OSPF (or ntpdate ran before OSPF is=20
brought up), which is no trouble if you have specialized hardware for=20
synching the clock. OTOH, the route to ntpd servers might not be up=20
before OSPF is run, and ntpd is completely incapable of handling=20
non-existent routes (it gets stuck forever on using an incorrect=20
interface, and has to be restarted).

Sorry for the rant. :-( I could never think of a solution to this=20
problem, and whenever I see a commit where I can see one wanting to do=20
things in the very opposite order... it upsets me. :-(

--=20
Daniel C. Sobral
Ger=EAncia de Opera=E7=F5es
Divis=E3o de Comunica=E7=E3o de Dados
Coordena=E7=E3o de Seguran=E7a
VIVO Centro Oeste Norte
Fones: 55-61-313-7654/Cel: 55-61-9618-0904
E-mail:	Daniel.Capo@tco.net.br
	Daniel.Sobral@tcoip.com.br
	dcs@tcoip.com.br




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?3FC72640.9060305>