Mailing List Archive

quagga and tun problems
hi!

I'm studying quagga and openvpn 2.0 on debian sarge. I'm using OSPF algorithm.

First of all I note that there are problems if openvpn sets client ip (peer
ptp is not tun ptp peer) and I have this error:
Packet from 10.8.0.x received on link tun0 but no ospf_interface

I solved by setting client ip explicitly.
Is a Quagga bug? Is there a new quagga version that fix it or another way to
solve it?

Now Quagga seems to work correctly but when I restart quagga daemons I have
these error messages:
"Please specify an existing interface"

How can I solve it?

Thanks very much,
Fabio


--

Dott. Fabio Marcone

2T srl
Telefono +39 - 0871- 540154
Fax +39 - 0871- 571594
Email fabio.marcone@duet.it
Indirizzo Viale B. Croce 573
66013 Chieti Scalo (CH)
GNU/Linux registered user #400424
_______________________________________________
Quagga-users mailing list
Quagga-users@lists.quagga.net
http://lists.quagga.net/mailman/listinfo/quagga-users
Re: quagga and tun problems [ In reply to ]
On Mon, 13 Feb 2006, Fabio wrote:

> Packet from 10.8.0.x received on link tun0 but no ospf_interface
>
> I solved by setting client ip explicitly. Is a Quagga bug?

Right now, no, don't think so. OpenVPNs' typical configuration seems
to be quite odd - check the list.

> Is there a new quagga version that fix it or another way to solve
> it?

There might some OpenVPN patches to make it behave more nicely:

http://www.google.ie/search?q=site%3Alists.quagga.net+openvpn

> Now Quagga seems to work correctly but when I restart quagga
> daemons I have these error messages:

> "Please specify an existing interface"

> How can I solve it?

You have a 'passive-interface ....' line in your config for which
there is no actual interface.

It really should be nicer about this, but it isn't. Please log an RFE
with bugzilla.

regards,
--
Paul Jakma paul@clubi.ie paul@jakma.org Key ID: 64A2FF6A
Fortune:
When in doubt, have a man come through the door with a gun in his hand.
-- Raymond Chandler
_______________________________________________
Quagga-users mailing list
Quagga-users@lists.quagga.net
http://lists.quagga.net/mailman/listinfo/quagga-users
Re: quagga and tun problems [ In reply to ]
Hi!

On Monday 13 February 2006 16:11, Paul Jakma wrote:
> On Mon, 13 Feb 2006, Fabio wrote:
> > Packet from 10.8.0.x received on link tun0 but no ospf_interface
> >
> > I solved by setting client ip explicitly. Is a Quagga bug?
>
> Right now, no, don't think so. OpenVPNs' typical configuration seems
> to be quite odd - check the list.
>
> > Is there a new quagga version that fix it or another way to solve
> > it?
>
> There might some OpenVPN patches to make it behave more nicely:
>
> http://www.google.ie/search?q=site%3Alists.quagga.net+openvpn
>
> > Now Quagga seems to work correctly but when I restart quagga
> > daemons I have these error messages:
> >
> > "Please specify an existing interface"
> >
> > How can I solve it?
>
> You have a 'passive-interface ....' line in your config for which
> there is no actual interface.
I should force quagga to use tun interface so I write passive-interface ethx
in ospd.conf. is it right?

Quagga howto isn't very clear about passive-interface parameter.

Thanks,
Fabio

>
> It really should be nicer about this, but it isn't. Please log an RFE
> with bugzilla.
>
> regards,

--

Dott. Fabio Marcone

2T srl
Telefono +39 - 0871- 540154
Fax +39 - 0871- 571594
Email fabio.marcone@duet.it
Indirizzo Viale B. Croce 573
66013 Chieti Scalo (CH)
GNU/Linux registered user #400424
_______________________________________________
Quagga-users mailing list
Quagga-users@lists.quagga.net
http://lists.quagga.net/mailman/listinfo/quagga-users
Re: quagga and tun problems [ In reply to ]
On Mon, 13 Feb 2006, Fabio wrote:

> I should force quagga to use tun interface so I write
> passive-interface ethx in ospd.conf. is it right?

Probably not ;)

> Quagga howto isn't very clear about passive-interface parameter.

Passive-interface is where you want an interface to not participate
in OSPF - ie ignore received packets. It will still be described in
the router-LSA though.

It's probably not what you need, but that depends on what problem you
were trying to solve.

regards,
--
Paul Jakma paul@clubi.ie paul@jakma.org Key ID: 64A2FF6A
Fortune:
Don't shout for help at night. You might wake your neighbors.
-- Stanislaw J. Lem, "Unkempt Thoughts"
_______________________________________________
Quagga-users mailing list
Quagga-users@lists.quagga.net
http://lists.quagga.net/mailman/listinfo/quagga-users
Re: quagga and tun problems [ In reply to ]
On Mon, Feb 13, 2006 at 03:11:30PM +0000, Paul Jakma wrote:
> On Mon, 13 Feb 2006, Fabio wrote:
> >"Please specify an existing interface"
>
> You have a 'passive-interface ....' line in your config for which
> there is no actual interface.
>
> It really should be nicer about this, but it isn't. Please log an RFE
> with bugzilla.

Is there any reason the warning message shouldn't give a bit
more info on the problem? Perhaps something like the
attached patch...

Regards,
Andy
Re: quagga and tun problems [ In reply to ]
On Mon, 13 Feb 2006, Andrew J. Schorr wrote:

> Is there any reason the warning message shouldn't give a bit more
> info on the problem? Perhaps something like the attached patch...

That's better, but really it should do what 'interface <foo>' does
and result in the interface config structures being setup (via the
if_new hook) - otherwise its effect is forgotten.

I.e. if this command was moved to interface node. We'd not have this
problem I think, ala:

interface eth0
passive-interface <optional address>

regards,
--
Paul Jakma paul@clubi.ie paul@jakma.org Key ID: 64A2FF6A
Fortune:
"Paul Lynde to block..."
-- a contestant on "Hollywood Squares"
_______________________________________________
Quagga-users mailing list
Quagga-users@lists.quagga.net
http://lists.quagga.net/mailman/listinfo/quagga-users
Re: quagga and tun problems [ In reply to ]
On Tue, Feb 14, 2006 at 06:43:17AM +0000, Paul Jakma wrote:
> On Mon, 13 Feb 2006, Andrew J. Schorr wrote:
>
> >Is there any reason the warning message shouldn't give a bit more
> >info on the problem? Perhaps something like the attached patch...
>
> That's better, but really it should do what 'interface <foo>' does
> and result in the interface config structures being setup (via the
> if_new hook) - otherwise its effect is forgotten.

Ah yes, I see.

> I.e. if this command was moved to interface node. We'd not have this
> problem I think, ala:
>
> interface eth0
> passive-interface <optional address>

Yes, that seems like a better solution.

It seems that ospf6d already has an "ipv6 ospf6 passive" interface
subcommand. And a quick web search indicates that an "ip ospf passive"
interface subcommand has been implemented by some vendors. I thought
Cisco had also done this, but I can't find it at the moment.

So I guess the question is whether this should be a general
interface subcommand or just an ospf subcommand. It should be pretty
easy to implement as an ospf interface subcommand...

Regards,
Andy
_______________________________________________
Quagga-users mailing list
Quagga-users@lists.quagga.net
http://lists.quagga.net/mailman/listinfo/quagga-users
Re: quagga and tun problems [ In reply to ]
On Tue, 14 Feb 2006, Andrew J. Schorr wrote:

> It seems that ospf6d already has an "ipv6 ospf6 passive" interface
> subcommand. And a quick web search indicates that an "ip ospf
> passive" interface subcommand has been implemented by some vendors.
> I thought Cisco had also done this, but I can't find it at the
> moment.

Nor can I. It would be nicer I think.

> So I guess the question is whether this should be a general
> interface subcommand or just an ospf subcommand. It should be
> pretty easy to implement as an ospf interface subcommand...

It would, but it'd be different.

Another might be to see whether the if_new_hook patch can be called,
as it would when you enter 'interface foo', where 'foo' does not
(yet)? exist.

regards,
--
Paul Jakma paul@clubi.ie paul@jakma.org Key ID: 64A2FF6A
Fortune:
"Hawk, we're going to die."
"Never say die... and certainly never say we."
-- M*A*S*H
_______________________________________________
Quagga-users mailing list
Quagga-users@lists.quagga.net
http://lists.quagga.net/mailman/listinfo/quagga-users
Re: quagga and tun problems [ In reply to ]
I know this is an old thread, but I was just bitten by it today.

On Tue, Feb 14, 2006 at 06:43:17AM +0000, Paul Jakma wrote:
> That's better, but really it should do what 'interface <foo>' does
> and result in the interface config structures being setup (via the
> if_new hook) - otherwise its effect is forgotten.

Can you see any reason why the attached (trivial) patch would not fix
the problem? It simply creates the interface (by calling
if_get_by_name instead of if_lookup_by_name) if it doesn't
already exist...

It seems to do the trick here.

Is it clear whether this should be done in both the "passive-interface"
and "no passive-interface" subcommands? I'm not clear on whether we
should instantiate the interface for a "no passive-interface" command.
It seems like this would be important if we ever support default
passive interfaces, but I'm not sure it matters at the moment...

Regards,
Andy
Re: quagga and tun problems [ In reply to ]
On Sat, 17 Jun 2006, Andrew J. Schorr wrote:

> I know this is an old thread, but I was just bitten by it today.

> Can you see any reason why the attached (trivial) patch would not
> fix the problem? It simply creates the interface (by calling
> if_get_by_name instead of if_lookup_by_name) if it doesn't already
> exist...
>
> It seems to do the trick here.

ACK.

> Is it clear whether this should be done in both the
> "passive-interface" and "no passive-interface" subcommands? I'm
> not clear on whether we should instantiate the interface for a "no
> passive-interface" command. It seems like this would be important
> if we ever support default passive interfaces, but I'm not sure it
> matters at the moment...

'reaping' interfaces that serve no purpose is ultimately probably
better handled as a seperate problem, rather than on a per-command
basis.

But right now, that could be left unchanged, yes.

regards,
--
Paul Jakma paul@clubi.ie paul@jakma.org Key ID: 64A2FF6A
Fortune:
I don't want to live on in my work, I want to live on in my apartment.
-- Woody Allen
_______________________________________________
Quagga-users mailing list
Quagga-users@lists.quagga.net
http://lists.quagga.net/mailman/listinfo/quagga-users
Re: quagga and tun problems [ In reply to ]
On Sun, Jun 18, 2006 at 12:53:11AM +0100, Paul Jakma wrote:
> On Sat, 17 Jun 2006, Andrew J. Schorr wrote:
> >Is it clear whether this should be done in both the
> >"passive-interface" and "no passive-interface" subcommands? I'm
> >not clear on whether we should instantiate the interface for a "no
> >passive-interface" command. It seems like this would be important
> >if we ever support default passive interfaces, but I'm not sure it
> >matters at the moment...
>
> 'reaping' interfaces that serve no purpose is ultimately probably
> better handled as a seperate problem, rather than on a per-command
> basis.

Agreed.

> But right now, that could be left unchanged, yes.

Hmmm, I can't quite make up my mind on this. I think my patch
is more elegant: just create the interface in either case.
But I recognize that it's not needed at the moment, since we
don't support passive by default. But if/when we do, we'll need to
handle this as in the patch (just create the interface). So perhaps
better to be prepared and not have lurking problems in the code?

On balance, I'm inclined to commit the patch as is. Does anyone feel strongly
that we should avoid creating the unneeded interface in the case
of a "no passive-interface" subcommand?

Regards,
Andy
_______________________________________________
Quagga-users mailing list
Quagga-users@lists.quagga.net
http://lists.quagga.net/mailman/listinfo/quagga-users
Re: quagga and tun problems [ In reply to ]
On Sat, 17 Jun 2006, Andrew J. Schorr wrote:

> Hmmm, I can't quite make up my mind on this. I think my patch is
> more elegant: just create the interface in either case.

Urg, apologies "could be left unchanged" - I didn't mean to suggest
it should /not/. ;)

> On balance, I'm inclined to commit the patch as is. Does anyone
> feel strongly that we should avoid creating the unneeded interface
> in the case of a "no passive-interface" subcommand?

No objections here at all. ACK on the patch :).

regards,
--
Paul Jakma paul@clubi.ie paul@jakma.org Key ID: 64A2FF6A
Fortune:
"Open the pod bay doors, HAL."
-- Dave Bowman, 2001
_______________________________________________
Quagga-users mailing list
Quagga-users@lists.quagga.net
http://lists.quagga.net/mailman/listinfo/quagga-users