mpls-linux-general — Mailing list for general discussion about MPLS for Linux

Re: [mpls-linux-general] problem in compilation

[James R. L. <jl...@mi...>]

On Tue, Apr 24, 2001 at 10:48:51AM -0700, Srinivas Reddy wrote:
> hi james,
> thanks for u r reply ,
> i will test it now ,but i am sure that i have not
> installed zebra,and i have one doubt is it necessary
> that i have to install total "zebra", or just i need
> to compile as you suggested 
> please advice me
> once again thaks for your reply
> regards
> lsreddy

If you want to compile ldp-portable with zebra support you will need the
entire zebra source.  You do not need to install it, you just need to compile
it.

You can always compile ldp_linux instead, it won't give you as much
functionality, but it works.

Jim

> --- "James R. Leu" <jl...@mi...> wrote:
> > Hello,
> > 
> > Doing a 'make ldp_zebra' from the top of
> > ldp-portable will give you
> > the disired result.  It will look for a 'zebra'
> > directory at the same level
> > as the 'ldp-portable' directory.  Make sure you have
> > configured and compiled
> > zebra first.  Also you will always want to 'cvs
> > update' your ldp-portable
> > tree often. (for example I fixed a compilation bug
> > just this morning).
> > 
> > 
> > [jleu jleu-laptop 6:51am]~/mpls/cvs-trees-> ls
> > ldp-portable/   mpls-linux/    zebra/
> > [jleu jleu-laptop 6:51am]~/mpls/cvs-trees-> cd
> > ldp-portable/
> > [jleu jleu-laptop
> > 6:52am]~/mpls/cvs-trees/ldp-portable-> make
> > ldp_zebra
> > 
> > 
> > On Tue, Apr 24, 2001 at 07:43:57AM +0530, L Srinivas
> > Reddy wrote:
> > > hi ,
> > > 	just now i downloaded mpls-linux and ldp-portable
> > from CVS repository,i
> > > have not installed full fledged "zebra".i am
> > getting the problem of missin
> > > zebra.h and etc..can some body give me idea where
> > can i get these zebra.h and
> > > so on.. and on more thing is no ware i found the
> > option -DMPLS_LINX in make
> > > files can i include this in make file . 
> > > out put of compilation of "zebra-ldp" 
> > > [root@lsreddy zebra-port]# make
> > > cc    -c -o crldp_path.o crldp_path.c
> > > crldp_path.c:3: zebra.h: No such file or directory
> > > crldp_path.c:5: command.h: No such file or
> > directory
> > > crldp_path.c:6: prefix.h: No such file or
> > directory
> > > crldp_path.c:7: linklist.h: No such file or
> > directory
> > > crldp_path.c:9: ldp_struct.h: No such file or
> > directory
> > > In file included from crldp_path.c:12:
> > > zldp_lib.h:7: zebra.h: No such file or directory
> > > zldp_lib.h:8: prefix.h: No such file or directory
> > > zldp_lib.h:10: zclient.h: No such file or
> > directory
> > > make: *** [crldp_path.o] Error 1
> > > thanks in advance
> > > expecting early reply
> > > regards
> > > lsreddy
> > > 
> > > _______________________________________________
> > > mpls-linux-general mailing list
> > > mpl...@li...
> > >
> >
> http://lists.sourceforge.net/lists/listinfo/mpls-linux-general
> > 
> > -- 
> > James R. Leu
> 
> 
> __________________________________________________
> Do You Yahoo!?
> Yahoo! Auctions - buy the things you want at great prices
> http://auctions.yahoo.com/

-- 
James R. Leu

Re: [mpls-linux-general] problem in compilation

[Srinivas R. <ls...@ya...>]

hi james,
thanks for u r reply ,
i will test it now ,but i am sure that i have not
installed zebra,and i have one doubt is it necessary
that i have to install total "zebra", or just i need
to compile as you suggested 
please advice me
once again thaks for your reply
regards
lsreddy
--- "James R. Leu" <jl...@mi...> wrote:
> Hello,
> 
> Doing a 'make ldp_zebra' from the top of
> ldp-portable will give you
> the disired result.  It will look for a 'zebra'
> directory at the same level
> as the 'ldp-portable' directory.  Make sure you have
> configured and compiled
> zebra first.  Also you will always want to 'cvs
> update' your ldp-portable
> tree often. (for example I fixed a compilation bug
> just this morning).
> 
> 
> [jleu jleu-laptop 6:51am]~/mpls/cvs-trees-> ls
> ldp-portable/   mpls-linux/    zebra/
> [jleu jleu-laptop 6:51am]~/mpls/cvs-trees-> cd
> ldp-portable/
> [jleu jleu-laptop
> 6:52am]~/mpls/cvs-trees/ldp-portable-> make
> ldp_zebra
> 
> 
> On Tue, Apr 24, 2001 at 07:43:57AM +0530, L Srinivas
> Reddy wrote:
> > hi ,
> > 	just now i downloaded mpls-linux and ldp-portable
> from CVS repository,i
> > have not installed full fledged "zebra".i am
> getting the problem of missin
> > zebra.h and etc..can some body give me idea where
> can i get these zebra.h and
> > so on.. and on more thing is no ware i found the
> option -DMPLS_LINX in make
> > files can i include this in make file . 
> > out put of compilation of "zebra-ldp" 
> > [root@lsreddy zebra-port]# make
> > cc    -c -o crldp_path.o crldp_path.c
> > crldp_path.c:3: zebra.h: No such file or directory
> > crldp_path.c:5: command.h: No such file or
> directory
> > crldp_path.c:6: prefix.h: No such file or
> directory
> > crldp_path.c:7: linklist.h: No such file or
> directory
> > crldp_path.c:9: ldp_struct.h: No such file or
> directory
> > In file included from crldp_path.c:12:
> > zldp_lib.h:7: zebra.h: No such file or directory
> > zldp_lib.h:8: prefix.h: No such file or directory
> > zldp_lib.h:10: zclient.h: No such file or
> directory
> > make: *** [crldp_path.o] Error 1
> > thanks in advance
> > expecting early reply
> > regards
> > lsreddy
> > 
> > _______________________________________________
> > mpls-linux-general mailing list
> > mpl...@li...
> >
>
http://lists.sourceforge.net/lists/listinfo/mpls-linux-general
> 
> -- 
> James R. Leu


__________________________________________________
Do You Yahoo!?
Yahoo! Auctions - buy the things you want at great prices
http://auctions.yahoo.com/

[mpls-linux-general] problem in compilation

[L S. R. <ls...@ya...>]

hi ,
	just now i downloaded mpls-linux and ldp-portable from CVS repository,i
have not installed full fledged "zebra".i am getting the problem of missin
zebra.h and etc..can some body give me idea where can i get these zebra.h and
so on.. and on more thing is no ware i found the option -DMPLS_LINX in make
files can i include this in make file . 
out put of compilation of "zebra-ldp" 
[root@lsreddy zebra-port]# make
cc    -c -o crldp_path.o crldp_path.c
crldp_path.c:3: zebra.h: No such file or directory
crldp_path.c:5: command.h: No such file or directory
crldp_path.c:6: prefix.h: No such file or directory
crldp_path.c:7: linklist.h: No such file or directory
crldp_path.c:9: ldp_struct.h: No such file or directory
In file included from crldp_path.c:12:
zldp_lib.h:7: zebra.h: No such file or directory
zldp_lib.h:8: prefix.h: No such file or directory
zldp_lib.h:10: zclient.h: No such file or directory
make: *** [crldp_path.o] Error 1
thanks in advance
expecting early reply
regards
lsreddy

Re: [mpls-linux-general] (no subject)

[James R. L. <jl...@mi...>]

On Fri, Apr 20, 2001 at 02:54:52PM +0200, Patrick Havel wrote:
> Hello everybody,
>=20
> my MPLS network now works using mplsadm (thank you James R. Leu).
> Now I want to add LDP.
> I have the following setup:
>=20
>  /----B----\
> A           D
>  \----C----/

First, the LDP spec doesn't support multipath.  ldp-portable makes an
attempt, but I would not have your first LDP testing include it :-)

Second, are you familiar with zebra? (www.zebra.org>
If so, grab the most recent version of zebra from CVS and grab the
most recent version of LDP portable from CVS and you can build
ldp-portable to run with zebra.  It's still very beta, but it works :-)

If not, you can still use ldp-portable (ldp_linux) you will just have
to use it's clunky interface.  Take a look at linux-port/README.cli for m=
ore
info on how to use it.  Make sure to grab the newest version from CVS!


> A's IP towards B xxx.xxx.2.10
> A's IP towards C xxx.xxx.4.10
>=20
> B's IP towards A xxx.xxx.2.20
> B's IP towards D xxx.xxx.3.10
>=20
> C's IP towards A xxx.xxx.4.20
> C's IP towards D xxx.xxx.5.10
>=20
> D's IP towards B xxx.xxx.3.20
> D's IP towards C xxx.xxx.5.20
>=20
> I have 4 LSP established from A to D via B (gen labels 10 for A to D an=
d 20
> for D to A) and A to D via C (gen labels 30 for A to D and 40 for D to =
A).
>=20
> All mashines have the LDP portable installed and it's running.
>=20
> My problem is, that I don't know how to configure the ldp protable
> correctly, so I'd be able to see something if I measure the traffic usi=
ng ethereal on
> a fifth mashine.
>=20
> Could anybody help me?
>=20
> Thnak you in advance.
>=20
> Patrick Havel
>=20
> --=20
> Ich freue mich immer(!) =FCber Eintr=E4ge in mein G=E4stebuch ;-)
>=20
> Homepage: http://www.informatik.uni-muenchen.de/~_havel
>=20
> GMX - Die Kommunikationsplattform im Internet.
> http://www.gmx.net
>=20
>=20
> _______________________________________________
> mpls-linux-general mailing list
> mpl...@li...
> http://lists.sourceforge.net/lists/listinfo/mpls-linux-general

--=20
James R. Leu

[mpls-linux-general] (no subject)

[Patrick H. <pat...@gm...>]

Hello everybody,

my MPLS network now works using mplsadm (thank you James R. Leu).
Now I want to add LDP.
I have the following setup:

 /----B----\
A           D
 \----C----/

A's IP towards B xxx.xxx.2.10
A's IP towards C xxx.xxx.4.10

B's IP towards A xxx.xxx.2.20
B's IP towards D xxx.xxx.3.10

C's IP towards A xxx.xxx.4.20
C's IP towards D xxx.xxx.5.10

D's IP towards B xxx.xxx.3.20
D's IP towards C xxx.xxx.5.20

I have 4 LSP established from A to D via B (gen labels 10 for A to D and 20
for D to A) and A to D via C (gen labels 30 for A to D and 40 for D to A).

All mashines have the LDP portable installed and it's running.

My problem is, that I don't know how to configure the ldp protable
correctly, so I'd be able to see something if I measure the traffic using ethereal on
a fifth mashine.

Could anybody help me?

Thnak you in advance.

Patrick Havel

-- 
Ich freue mich immer(!) über Einträge in mein Gästebuch ;-)

Homepage: http://www.informatik.uni-muenchen.de/~_havel

GMX - Die Kommunikationsplattform im Internet.
http://www.gmx.net

[mpls-linux-general] ldp-portable

[<Jac...@sw...>]

Dear all,


Is it possible to use ldp -portable with the "gated" routing code instead of
the "zebra"one.

Thank you for an answer,


Jacques

[mpls-linux-general] subscribe

[Victorianus K. <Vic...@ms...>]

subscribe

[mpls-linux-general] Capturing MPLS packets on CLIP interface

[Jonathan E. <je...@no...>]

Hi,

I've been trying to modify the latest tcpdump to read MPLS packets off ATM
interfaces, and I've noticed that in pcap_read_packet(), the function never
receives a packet from the kernel (via recvfrom() ) if it's been put onto an
LSP.  IP works fine.

Anything I'm missing?

Cheers!
Jon

[mpls-linux-general] mplsroute sample config

[Joern S. <j....@gm...>]

Hi everybody,

some people asked me for the static-routes-program. Due to the change of my 
workingplace, I forgot to include a sample config file! I apologize for this 
and send it this way.

Hope, this is useful for you.

Jörn



###################################################################
#
# STATIC ROUTES FOR MPLS
#
#
###################################################################

#
# default definitions should be placed here
#
DEFAULT_BEGIN
key gen
ipversion ipv4
DEFAULT_END

#
# definition of incoming labels
#
# using the form:
# label          interfacees

#INCOMING_BEGIN
# this one is only for testing
#1024    eth2
#1022    eth1
#INCOMING_END

#
# definition of outgoing labels with tunnel
#
# using the form:
# tunnel-name    label   interface       local address   next hop
#OUTGOING_BEGIN
#mpls0           1024    eth0            192.168.4.204   192.168.4.203
#OUTGOING_END

#
# definition for a mpls-switching router
#
# using the form:
# from interfaces  from label    to interface    to label   to local addr  to 
next hop
SWITCHING_BEGIN
eth2     1024    eth0    1023    192.168.4.203   192.168.4.203
eth0     1022    eth2    1021    192.168.3.204   192.168.3.201
SWITCHING_END

[mpls-linux-general] mpls w/ cisco router

[James R. <jr...@ci...>]

Hello,

I am using mpls 0.800 w/ ldp-portable on a 2.4.0 linux machine. I am =
trying to set it up as a LER. This LER will have two interfaces, one =
going to a single PC, and the other interface as a next hop being a =
Cisco7206 VXR router. Has anyone else configured a similar type of =
setup?

Thanks,
Jim Rowe

Re: [mpls-linux-general] MPLS on gigabit ethernet

[James R. L. <jl...@mi...>]

On Mon, Apr 16, 2001 at 10:23:06AM -0400, Jonathan Earle wrote:
> 
> Anyone tried this yet?  Any issues?

The Linux kernel treats gige and 10/100 as the same.  So there should not
be a problem (famous last words).

Jim
> 
> Jon
> 
> _______________________________________________
> mpls-linux-general mailing list
> mpl...@li...
> http://lists.sourceforge.net/lists/listinfo/mpls-linux-general

-- 
James R. Leu

[mpls-linux-general] MPLS on gigabit ethernet

[Jonathan E. <je...@no...>]

Anyone tried this yet?  Any issues?

Jon

Re: [mpls-linux-general] ldp question

[James R. L. <jl...@mi...>]

I trued to send a reponse to this a week ago, but I got a delivery error,
hopefully this one makes it to you.


Date: Fri, 6 Apr 2001 13:37:25 -0500
From: "James R. Leu" <jl...@mi...>
To: Scott Lee <nat...@ya...>
Cc: jl...@mi..., uk...@rz...
Subject: Re: ldp question

Hello,

Sorry for not getting back sooner.  Yon's answer reminded me about this.

On Thu, Apr 05, 2001 at 09:59:49AM -0700, Scott Lee wrote:
> Hi James,
> 
> My name is Scott Lee and I am one of many who
> is interested in Linux MPLS, LDP, transporting layer
> 2 over IP backbone, etc.
> 
> Can  I ask you a quick question?
> 
> When operating in downstream-unsolicited, liberal
> retention mode, 
> independent control mode, do you ever issue Label
> Request Message?

If you (the LSR) do not support label merge and have more then one
upstream neighbor to whom you must propogate an unsolicited
advertisment that you received from downstream, then you will need to
request more labels to satisfy building these LSPs.  Whether you propogate
a mapping and then request or request and then propogate a mapping is
determined by your control mode.

How sane is it to run unsolicited and not support label merge?  It's
not, BUT if your running over ethernet and you and your peer disagree
about distribution mode the RFC specifies that DU will be used.  SO ...
it is very possible that you will be in this situation and have no control
over it.

> According to the MPLS architecture RFC, the above
> mode should lead to the following scheme.
> {pushUnconditional, RequestNever, NoReleaseOnChange,
> noLoopDetection}
> 
> It seems like we never send label request message.
> 
> One more question, when you send label withdraw
> messsage and if you don't get label release message,
> what do you do?

I think Yon brought up a good point that a timer is needed, but it
doesn't seem like the RFC specifies one.  I know I don't handle this
case yet :-)  I'll add it to my list.


> 
> Thanks!
> 
> 
> 
> 
> __________________________________________________
> Do You Yahoo!?
> Get email at your own domain with Yahoo! Mail. 
> http://personal.mail.yahoo.com/

-- 
James R. Leu

On Thu, Apr 12, 2001 at 03:31:14PM -0700, Scott Lee wrote:
> 
> When operating in downstream-unsolicited, liberal
> retention mode, 
> independent control mode, do you ever issue Label
> Request Message? According to MPLS architecture doc,
> it seems like LDP would never send Label Request msg.
> 
> One more question.
> When you send label withdraw message and you don't get
> label release message, what do you do in that case?
> Should LDP timeout and resend the withdraw message?
> 
> Thanks.
> 
> 
> 
> 
> 
> __________________________________________________
> Do You Yahoo!?
> Get email at your own domain with Yahoo! Mail. 
> http://personal.mail.yahoo.com/
> 
> _______________________________________________
> mpls-linux-general mailing list
> mpl...@li...
> http://lists.sourceforge.net/lists/listinfo/mpls-linux-general

-- 
James R. Leu

[mpls-linux-general] ldp question

[Scott L. <nat...@ya...>]

When operating in downstream-unsolicited, liberal
retention mode, 
independent control mode, do you ever issue Label
Request Message? According to MPLS architecture doc,
it seems like LDP would never send Label Request msg.

One more question.
When you send label withdraw message and you don't get
label release message, what do you do in that case?
Should LDP timeout and resend the withdraw message?

Thanks.





__________________________________________________
Do You Yahoo!?
Get email at your own domain with Yahoo! Mail. 
http://personal.mail.yahoo.com/

[mpls-linux-general] MPLS on gigabit ethernet

[Jonathan E. <je...@no...>]

Anyone tried this yet?  Any issues?

Jon

Re: [mpls-linux-general] (no subject)

[James R. L. <jl...@mi...>]

On Tue, Apr 10, 2001 at 12:10:14PM +0200, Patrick Havel wrote:
> Hello everybody,
>=20
> I want to create a LSP within 3 computers.
> A------B-------C
>=20
> A's IP: 192.168.2.1, eth0
> B's IP towards A: 192.168.2.2, eth1
> B's IP towards C: 192.168.3.1, eth0
> C's IP: 192.168.3.2, eth1

Remeber LSPs are uni-directional. TO have trafffic sent and recieved
on an LSP you need to build two.  In your case LSP "16" will be used
to send from from A to C.  LSP "17" will be used to sent traffic from C t=
o A.

LSP from A -> C
---------------

A: route add -host 192.168.3.2 gw 192.168.2.2
A: mplsadm -A -B -O gen:16:eth0:ipv4:192.168.2.2 -f 192.168.3.2/32

B: mplsadm -L eth1:0
B: mplsadm -A -B -I gen:16:0 -O gen:16:eth0:ipv4:192.168.3.2

C: mplsadm -L eth1:0
C: mplsadm -A -I gen:16:0

LSP from C -> A
---------------

C: route add -host 192.168.2.1 gw 192.168.3.1
C: mplsadm -A -B -O gen:17:eth1:ipv4:192.168.3.1 -f 192.168.2.1/32

B: mplsadm -L eth0:0
B: mplsadm -A -B -I gen:17:0 -O gen:17:eth1:ipv4:192.168.2.1

A: mplsadm -L eth0:0
A: mplsadm -A -I gen:17:0

> I want to meassure the traffic using ethereal on mashine A or C.=20
> In the next step I'd like to include LDP also. If this works I want=20
> to add a 4th computer parallel to B and do some meassuring.

NOTE: Ethereal and tcpdump will show "weird" packet when running ON the
LER (A or C in this setup).  This is due to where (in the Linux net stack=
)
tcpdump and ethereal snag a copy of the packet.  Right now they get a ver=
sion
of the packet that has been partially modified.

For LDP, make sure to grab ldp-portable from CVS to get the latest versio=
n.

> I hope somebody can give me some advice.
>=20
> Thank you in advance.
>=20
> Patrick Havel
>=20
> --=20
> Ich freue mich immer(!) =FCber Eintr=E4ge in mein G=E4stebuch ;-)
>=20
> Homepage: http://www.informatik.uni-muenchen.de/~_havel
>=20
> GMX - Die Kommunikationsplattform im Internet.
> http://www.gmx.net
>=20
>=20
> _______________________________________________
> mpls-linux-general mailing list
> mpl...@li...
> http://lists.sourceforge.net/lists/listinfo/mpls-linux-general

--=20
James R. Leu

[mpls-linux-general] (no subject)

[Patrick H. <pat...@gm...>]

Hello everybody,

I want to create a LSP within 3 computers.
A------B-------C

A's IP: 192.168.2.1, eth0
B's IP towards A: 192.168.2.2, eth1
B's IP towards C: 192.168.3.1, eth0
C's IP: 192.168.3.2, eth1

On A:
route add -host 192.168.2.2 gw 192.168.2.2
mplsadm -A -O gen:16:eth0:ipv4:192.168.2.2
mplsadm -B -O gen:16:eth0 -f 192.168.2.2/32
mplsadm -L eth0:0

On C:
route add -host 192.168.3.1 gw 192.168.3.1
mplsadm -A -O gen:17:eth1:ipv4:192.168.3.1
mplsadm -B -O gen:17:eth1 -f 192.168.3.1/32
mplsadm -L eth1:0

On B:
mplsadm -A -B -I gen:16:eth1
mplsadm -A -B -I gen:17:eth0

I want to meassure the traffic using ethereal on mashine A or C. 
In the next step I'd like to include LDP also. If this works I want 
to add a 4th computer parallel to B and do some meassuring.

I hope somebody can give me some advice.

Thank you in advance.

Patrick Havel

-- 
Ich freue mich immer(!) über Einträge in mein Gästebuch ;-)

Homepage: http://www.informatik.uni-muenchen.de/~_havel

GMX - Die Kommunikationsplattform im Internet.
http://www.gmx.net

RE: [mpls-linux-general] VPNs using MPLS-Linux/LDP-Portable

[Yon U. <uk...@rz...>]

Hello,

 excuse my ignorance, is Nanda or Debasis your family name (and where do
you come from?)?


On Thu, 5 Apr 2001, Nanda, Debasis wrote:
> Hi Yon,
> This is really great . I did work on Leu's MPLS package and found it very
> useful . I am planning to integrate this with the zldpd . just had couple of
> questions before I start.
>
> 1) Will I need LDP portable if I am using zldpd ?

 YES, zldpd is based on ldp-portable. It is just the beginning of a port
to the zebra framework of ldp-portable.

> 2) What are the points og attachment between ldpd+Zebra with mpls forwarding
> plane .

 At the moment, only the file zldp_mpls_impl.c contains the (empty)
adapter to the mpls forwarding plane. You could copy the ldp_mpls_impl.c
file from the linux port and expect some kind of (half-broken)
functionality. The API is declared in the file ldp_mpls_impl.h, part of
ldp-portable.


> 3) Does zldpd has traffic extentions (TE) implemented ?

TE for ldp? What extensions do you exactly refer to? ldp-portable includes
a cr-ldp implementation (AFAIK, ask Jim), but I've yet to incorporate
it.

> 4) What will be the impact of running the BGP4 daemon simultaneously with
> zldpd ?

 Hmm, with respect to what? If the machine is adequately equipped, there
shall be no contention for resources (cpu/mem/...). Of course, you might
want something more useful out of the combination ;) for example bgp/mpls
VPNs. This is FFS as it implies some changes to the zebra daemon. At the
moment, and assuming you get it working with the linux port forwarding
plane adapter, you can expect some lsp forwarding. Nothing special about
BGP just, probably, as you wont be announcing tags for all of the
internet routes, you wont get no benefits out of the combination. The main
zebra daemon should be informed of the existance of a lsp for the bgp
next-hop(s) some edge router is announcing and it could bind the routes
matching the next-hop(s) to the lsp. This would be a poor man's
get-full-routes-out-of-core solution. As is, it will just bind to
explicitely announced routes (FECs) and will do no recursion/tunneling.

> I am also looking at an RSVP implementation with TE extentions on Linux to
> interoperate with Jame's MPLS forwarding plane .

 If you find some open source (even gpl) rsvp-te implementation, let me
know, please.


> Anticipating a quick response.

Here it goes.

> Regards
> Debasis Nanda
> NEC America

sincerely yours,
 yon

YON international
CEO, CTO, CFO, VP of everything



> -----Original Message-----
> From: Yon Uriarte [mailto:uk...@rz...]
> Sent: Wednesday, February 21, 2001 11:27 AM
> To: James R. Leu
> Cc: MPLS-Linux
> Subject: Re: [mpls-linux-general] VPNs using MPLS-Linux/LDP-Portable
>
>
> On Tue, 20 Feb 2001, James R. Leu wrote:
>
> > >  I have started on 3, it is a mess, doesn't do anything useful
> > > yet, as I have problems understanding ldp-portable (to be
> > > precise: I don't want to learn how ldp-portable works,
> > > need-to-know is my keyword here).
> >
> > This would be great!  How can I help.  I can explain any of the
> ldp-portable
> > API for you, what needs to be provided to ldp-portable, and how things
> > need to be stored.
>
>  Well, as things go on, I'll have more questions. I've put the little
> monster up for grabbing at:
>
> http://wwwstud.informatik.uni-karlsruhe.de/~s_uriart/zldpd.tgz
>
>  Take a look at it, if you want to. I guess it could be made to run
> with direct mpls-linux access, but everything else running over
> zebra.
>
>
>  I haven't quite tested it. It compiles, and runs a bit without
> crashing. If you don't start zebra before starting it, it will
> crash (on an assert). No need to run zebra as root, btw, just
> start it, so zldpd will know about interfaces (and later routes).
>
>
> > I have time to answer question, just not time to do the port
> > myself :-)
>
>  Fine! As it goes on, I'll ask you precise questions. At the
> moment I just wanted to get the ball rolling and take a look
> at the mpls/ldp thingie. Your code looks really professional.
>
>
>
> enjoy & HAND,
>  yon
>
>
> PS: forgot to put a sample configuration file in the tgz.
>   I attach it.
>
>

-- 
 For the traditional victims, the New World Order is not likely to
 be an improvement on the old, [...] .
                                     Noam Chomsky

[mpls-linux-general] information required

[L S. R. <ls...@ya...>]

hi all,
 	i want information that is there any thing required other than linux
kernel 2.4.x and mpls patch for mpls to install,can some give me some URL which
will have most valueble information about this project.i am new to this mpls
please help me.
thanks in advance 
regards
lsrbsd

Re: [mpls-linux-general] MPLS and ATM

[James R. L. <jl...@mi...>]

Hello,

This is great debugging info!

>   mpls_atm_create_vcc: vci 33 <-> 1024 vpi 0 <-> 1

Here we see where I consult the card VPI/VCI ranges

>   atm_connect (TX: cl 0,bw 0-0,sdu 65535; RX: cl 1,bw 0-0,sdu 65535,AAL 5)

Here we see the params I sent to the ATM layer, notice "sdu 65535".

>   Apr  5 16:06:37 onc2 kernel: IA:  SDU size over the configured SDU size
> 10240

IA doesn't like an sdu 65535 I guess.  So I think the correct fix for this is
going to be to consult with the card for an SDU size instead of just blindly
using 65535.

Thanks for the great debugging info!

Jim

> >>EIP; d083daa3 <[iphase]ia_close+1b3/330>   <=====
> Trace; d083dca9 <[iphase]ia_open+89/d8>
> Trace; d083dce1 <[iphase]ia_open+c1/d8>
> Trace; c01edd70 <atm_do_connect_dev+14c/184>
> Trace; c01eddd6 <atm_do_connect+2e/34>
> Trace; c01ede9d <atm_connect_vcc+c1/120>
> Trace; c01dfa36 <mpls_atm_create_vcc+1ca/244>
> Trace; c01dcee7 <mpls_add_out_label+20b/4ac>
> Trace; c014f486 <nfs_read_rpcsetup+3a/ec>
> Trace; c01dc599 <mpls_rtm_newnhlfe+d/14>
> Trace; c019e269 <rtnetlink_rcv+24d/38c>
> Trace; c01e0a2e <netlink_data_ready+1a/60>
> Trace; c01e0436 <netlink_unicast+296/2dc>
> Trace; c01e08b7 <netlink_sendmsg+1df/1f0>
> Trace; c01e06d8 <netlink_sendmsg+0/1f0>
> Trace; c019508d <sock_sendmsg+81/a4>
> Trace; c01e06d8 <netlink_sendmsg+0/1f0>
> Trace; c0195cfc <sys_sendto+d0/f0>
> Trace; c01b16c8 <ip_rcv+358/390>
> Trace; c01b17f0 <ip_rcv_finish+0/210>
> Trace; d083d0f5 <[iphase]ia_int+85/94>
> Trace; c019a1cd <net_rx_action+125/240>
> Trace; c01964cb <sys_socketcall+123/1dc>
> Trace; c0106daf <system_call+33/38>
> Code;  d083daa3 <[iphase]ia_close+1b3/330>
> 00000000 <_EIP>:
> Code;  d083daa3 <[iphase]ia_close+1b3/330>   <=====
>    0:   f7 77 0c                  div    0xc(%edi),%eax   <=====
> Code;  d083daa6 <[iphase]ia_close+1b6/330>
>    3:   89 c3                     mov    %eax,%ebx
> Code;  d083daa8 <[iphase]ia_close+1b8/330>
>    5:   b8 01 00 00 00            mov    $0x1,%eax
> Code;  d083daad <[iphase]ia_close+1bd/330>
>    a:   85 db                     test   %ebx,%ebx
> Code;  d083daaf <[iphase]ia_close+1bf/330>
>    c:   0f 44 d8                  cmove  %eax,%ebx
> Code;  d083dab2 <[iphase]ia_close+1c2/330>
>    f:   80 7f 2c 00               cmpb   $0x0,0x2c(%edi)
> Code;  d083dab6 <[iphase]ia_close+1c6/330>
>   13:   74 00                     je     15 <_EIP+0x15> d083dab8
> <[iphase]ia_close+1c8/330>
> 
> > -----Original Message-----
> > From: James R. Leu [mailto:jl...@mi...]
> > Sent: Wednesday, April 04, 2001 8:27 PM
> > To: Earle, Jonathan [CAR:1A31:EXCH]
> > Cc: 'Linux MPLS List'
> > Subject: Re: [mpls-linux-general] MPLS and ATM
> > 
> > 
> > Hello,
> > 
> > What is the atm index of your ATM card?
> > 
> > cat /proc/net/atm/device
> > 
> > It shows up as the Intf.
> > 
> > When creating your atm interface (via atmarp -c) the index in the name
> > needs to match that of the index from above.
> > 
> > Example:
> > 
> > If the atm card index (from /proc/net/atm/device) is 3
> > 
> > atmarp -c atm3
> > 
> > I know this is a kind of a ridiculous rule, but due to the nature of
> > ATM for Linux it is a requirement.  Someday I'll get a around 
> > to re-writing
> > the way ATM interfaces are handled on Linux ...
> > 
> > Let me know if this still doesn't help.
> > 
> > Jim
> > 
> > On Wed, Apr 04, 2001 at 05:51:21PM -0400, Jonathan Earle wrote:
> > > Hi James,
> > > 
> > > Setup is as follows:
> > > 
> > > Interphase 5575 ATM NIC
> > > Kernel 2.4.3
> > > MPLS v0.990
> > > 
> > > Basic IP connectivity between the two NICs works fine.  I 
> > configure the NICs
> > > thusly:
> > > 
> > > master)
> > >        atmsigd -b -m network
> > >        atmarpd -b
> > >        sleep 2s
> > > 
> > >        if [ "$adapter" != "atm$anum" ]; then
> > >           atmarp -c atm$anum
> > >        fi
> > > 
> > >        ifconfig atm$anum 192.168.1.1 up
> > >        sleep 2s
> > >        atmarp -s 192.168.1.2 0.0.102
> > >        ;;
> > >  slave)
> > >        atmsigd -b
> > >        atmarpd -b
> > >        sleep 2s
> > > 
> > >        if [ "$adapter" != "atm$anum" ]; then
> > >           atmarp -c atm$anum
> > >        fi
> > > 
> > >        ifconfig atm$anum 192.168.1.2 up
> > >        sleep 2s
> > >        atmarp -s 192.168.1.1 0.0.102
> > >        ;;
> > > 
> > > Using the examples in REAME.atm, I can create the labelspace:
> > > 
> > > root@onc4:~> cat /proc/net/mpls_labelspace
> > > lo      0
> > > teql0   0
> > > atm1    3
> > > 
> > > But not IN labels:
> > > 
> > > mpls_add_in_label: enter
> > > mpls_info_default_in_instruction: enter
> > > mpls_info_default_in_instruction: exit
> > > mpls_atm_create_vcc: enter(1,0,33,1)
> > > mpls_atm_create_vcc: adev == NULL
> > > mpls_add_in_label: failed creating vcc
> > > __mpls_del_in_label: enter
> > > __mpls_del_in_label: exit
> > > mpls_add_in_label: exit
> > > 
> > > or OUT labels:
> > > 
> > > mpls_add_out_label: enter
> > > mpls_info_default_moi_instruction: enter
> > > mpls_info_default_moi_instruction: exit
> > > mpls_atm_create_vcc: enter(1,0,33,2)
> > > mpls_atm_create_vcc: adev == NULL
> > > mpls_add_out_label: failed to create VCC
> > > __mpls_del_out_label: enter
> > > __mpls_del_out_label: exit
> > > mpls_add_out_label: exit
> > > 
> > > Cheers!
> > > Jon
> > > 
> > > _______________________________________________
> > > mpls-linux-general mailing list
> > > mpl...@li...
> > > http://lists.sourceforge.net/lists/listinfo/mpls-linux-general
> > 
> > -- 
> > James R. Leu
> > 

-- 
James R. Leu

RE: [mpls-linux-general] MPLS and ATM

[Jonathan E. <je...@no...>]

Hi Jim,

Okay.  I was taking your example a bit literally.  I'm now using atm0, not
atm1, since:

  root@onc4:~> cat /proc/net/atm/devices
  Itf Type    ESI/"MAC"addr AAL(TX,err,RX,err,drop) ...
    0 ia      00007794577c  0 ( 0 0 0 0 0 )  5 ( 273 0 93 0 0 )

shows the card is at index 0.  Okay.  On the target, I can create a new
labelspace and inbound label (I saved the 0.990 vesion of mplsadm as
mplsadm.new to avoid conflict with the previous version of the tool in a
shared tools directory):

  root@onc4:~> mplsadm.new -L atm0:3
  root@onc4:~> mplsadm.new -A -I atm:0/33:3

  root@onc4:~> cat /proc/net/mpls_labelspace
  lo      0
  teql0   0
  atm0    3

  root@onc4:~> cat /proc/net/mpls_in
  80210003 atm 0/33 3 POP DLV

debug output looks like:

  mpls_add_in_label: enter
  mpls_info_default_in_instruction: enter
  mpls_info_default_in_instruction: exit
  mpls_atm_create_vcc: enter(0,0,33,1)
  mpls_atm_create_vcc: vci 33 <-> 1024 vpi 0 <-> 1
  mpls_atm_create_vcc: create_vcc
  mpls_atm_create_vcc: setting up vcc
  atm_connect (TX: cl 0,bw 0-0,sdu 65535; RX: cl 1,bw 0-0,sdu 65535,AAL 5)
  mpls_atm_create_vcc: exit
  mpls_add_in_label: succeeded creating vcc
  mpls_add_in_label: exit

So far, so good.  Now, on to the source.  First, lets create the outbound
label:

  root@onc2:~> mplsadm.new -A -O atm:0/33:atm0
  Segmentation fault

Oops!  The log showed this right at the top of the oops output:

  Apr  5 16:06:37 onc2 kernel: IA:  SDU size over the configured SDU size
10240

IA is the Interphase adapter.

The processed oops output is:

ksymoops 0.7c on i686 2.4.3.  Options used
     -V (default)
     -k /proc/ksyms (default)
     -l /proc/modules (default)
     -o /lib/modules/2.4.3/ (default)
     -m /usr/src/linux/System.map (default)

Warning: You did not tell me where to find symbol information.  I will
assume that the log matches the kernel and modules that are running
right now and I'll use the default options above for symbol resolution.
If the current kernel and/or modules do not match the log, you can get
more accurate output by telling me the kernel version and where to find
map, modules, ksyms etc.  ksymoops -h explains the options.

CPU:    0
EIP:    0010:[<d083daa3>]
Using defaults from ksymoops -t elf32-i386 -a i386
EFLAGS: 00010046
eax: 000493e0   ebx: ced80044   ecx: 00000000   edx: 00000000
esi: ced80000   edi: ced51260   ebp: ced3e000   esp: cedc9c48
ds: 0018   es: 0018   ss: 0018
Process mplsadm.new (pid: 833, stackpage=cedc9000)
Stack: ffffffea ced3e000 cfa0a680 00000021 00035a18 00000286 ced51260
cedc9c64
       cedc9c64 00000000 00000246 cedc9c74 cedc9c74 00000000 d083dca9
d083dce1
       ced3e000 00000000 ced3e000 00000000 c01edd70 ced3e000 00000000
00000021
Call Trace: [<d083dca9>] [<d083dce1>] [<c01edd70>] [<c01eddd6>] [<c01ede9d>]
[<c01dfa36>] [<c01dcee7>]
       [<c014f486>] [<c01dc599>] [<c019e269>] [<c01e0a2e>] [<c01e0436>]
[<c01e08b7>] [<c01e06d8>] [<c019508d>]
       [<c01e06d8>] [<c0195cfc>] [<c01b16c8>] [<c01b17f0>] [<d083d0f5>]
[<c019a1cd>] [<c01964cb>] [<c0106daf>]
Code: f7 77 0c 89 c3 b8 01 00 00 00 85 db 0f 44 d8 80 7f 2c 00 74

>>EIP; d083daa3 <[iphase]ia_close+1b3/330>   <=====
Trace; d083dca9 <[iphase]ia_open+89/d8>
Trace; d083dce1 <[iphase]ia_open+c1/d8>
Trace; c01edd70 <atm_do_connect_dev+14c/184>
Trace; c01eddd6 <atm_do_connect+2e/34>
Trace; c01ede9d <atm_connect_vcc+c1/120>
Trace; c01dfa36 <mpls_atm_create_vcc+1ca/244>
Trace; c01dcee7 <mpls_add_out_label+20b/4ac>
Trace; c014f486 <nfs_read_rpcsetup+3a/ec>
Trace; c01dc599 <mpls_rtm_newnhlfe+d/14>
Trace; c019e269 <rtnetlink_rcv+24d/38c>
Trace; c01e0a2e <netlink_data_ready+1a/60>
Trace; c01e0436 <netlink_unicast+296/2dc>
Trace; c01e08b7 <netlink_sendmsg+1df/1f0>
Trace; c01e06d8 <netlink_sendmsg+0/1f0>
Trace; c019508d <sock_sendmsg+81/a4>
Trace; c01e06d8 <netlink_sendmsg+0/1f0>
Trace; c0195cfc <sys_sendto+d0/f0>
Trace; c01b16c8 <ip_rcv+358/390>
Trace; c01b17f0 <ip_rcv_finish+0/210>
Trace; d083d0f5 <[iphase]ia_int+85/94>
Trace; c019a1cd <net_rx_action+125/240>
Trace; c01964cb <sys_socketcall+123/1dc>
Trace; c0106daf <system_call+33/38>
Code;  d083daa3 <[iphase]ia_close+1b3/330>
00000000 <_EIP>:
Code;  d083daa3 <[iphase]ia_close+1b3/330>   <=====
   0:   f7 77 0c                  div    0xc(%edi),%eax   <=====
Code;  d083daa6 <[iphase]ia_close+1b6/330>
   3:   89 c3                     mov    %eax,%ebx
Code;  d083daa8 <[iphase]ia_close+1b8/330>
   5:   b8 01 00 00 00            mov    $0x1,%eax
Code;  d083daad <[iphase]ia_close+1bd/330>
   a:   85 db                     test   %ebx,%ebx
Code;  d083daaf <[iphase]ia_close+1bf/330>
   c:   0f 44 d8                  cmove  %eax,%ebx
Code;  d083dab2 <[iphase]ia_close+1c2/330>
   f:   80 7f 2c 00               cmpb   $0x0,0x2c(%edi)
Code;  d083dab6 <[iphase]ia_close+1c6/330>
  13:   74 00                     je     15 <_EIP+0x15> d083dab8
<[iphase]ia_close+1c8/330>

1 warning issued.  Results may not be reliable.

Cheers!
Jon


> -----Original Message-----
> From: James R. Leu [mailto:jl...@mi...]
> Sent: Wednesday, April 04, 2001 8:27 PM
> To: Earle, Jonathan [CAR:1A31:EXCH]
> Cc: 'Linux MPLS List'
> Subject: Re: [mpls-linux-general] MPLS and ATM
> 
> 
> Hello,
> 
> What is the atm index of your ATM card?
> 
> cat /proc/net/atm/device
> 
> It shows up as the Intf.
> 
> When creating your atm interface (via atmarp -c) the index in the name
> needs to match that of the index from above.
> 
> Example:
> 
> If the atm card index (from /proc/net/atm/device) is 3
> 
> atmarp -c atm3
> 
> I know this is a kind of a ridiculous rule, but due to the nature of
> ATM for Linux it is a requirement.  Someday I'll get a around 
> to re-writing
> the way ATM interfaces are handled on Linux ...
> 
> Let me know if this still doesn't help.
> 
> Jim
> 
> On Wed, Apr 04, 2001 at 05:51:21PM -0400, Jonathan Earle wrote:
> > Hi James,
> > 
> > Setup is as follows:
> > 
> > Interphase 5575 ATM NIC
> > Kernel 2.4.3
> > MPLS v0.990
> > 
> > Basic IP connectivity between the two NICs works fine.  I 
> configure the NICs
> > thusly:
> > 
> > master)
> >        atmsigd -b -m network
> >        atmarpd -b
> >        sleep 2s
> > 
> >        if [ "$adapter" != "atm$anum" ]; then
> >           atmarp -c atm$anum
> >        fi
> > 
> >        ifconfig atm$anum 192.168.1.1 up
> >        sleep 2s
> >        atmarp -s 192.168.1.2 0.0.102
> >        ;;
> >  slave)
> >        atmsigd -b
> >        atmarpd -b
> >        sleep 2s
> > 
> >        if [ "$adapter" != "atm$anum" ]; then
> >           atmarp -c atm$anum
> >        fi
> > 
> >        ifconfig atm$anum 192.168.1.2 up
> >        sleep 2s
> >        atmarp -s 192.168.1.1 0.0.102
> >        ;;
> > 
> > Using the examples in REAME.atm, I can create the labelspace:
> > 
> > root@onc4:~> cat /proc/net/mpls_labelspace
> > lo      0
> > teql0   0
> > atm1    3
> > 
> > But not IN labels:
> > 
> > mpls_add_in_label: enter
> > mpls_info_default_in_instruction: enter
> > mpls_info_default_in_instruction: exit
> > mpls_atm_create_vcc: enter(1,0,33,1)
> > mpls_atm_create_vcc: adev == NULL
> > mpls_add_in_label: failed creating vcc
> > __mpls_del_in_label: enter
> > __mpls_del_in_label: exit
> > mpls_add_in_label: exit
> > 
> > or OUT labels:
> > 
> > mpls_add_out_label: enter
> > mpls_info_default_moi_instruction: enter
> > mpls_info_default_moi_instruction: exit
> > mpls_atm_create_vcc: enter(1,0,33,2)
> > mpls_atm_create_vcc: adev == NULL
> > mpls_add_out_label: failed to create VCC
> > __mpls_del_out_label: enter
> > __mpls_del_out_label: exit
> > mpls_add_out_label: exit
> > 
> > Cheers!
> > Jon
> > 
> > _______________________________________________
> > mpls-linux-general mailing list
> > mpl...@li...
> > http://lists.sourceforge.net/lists/listinfo/mpls-linux-general
> 
> -- 
> James R. Leu
> 

[mpls-linux-general] Kernel Panic!!!

[Karthikeyan N. <ntk...@it...>]

hi,
    i tried to setup an LSP between two machines. I followed the architecture
given in utils. directory.(README.example). I was able to set up the LSP
properly. When i tried to send some tcp traffic using ttcp through the LSP
kernel panics and the system hangs. 
      From the call trace, i found out that it panics somewhere between 
"skb_under_panic"  and "alloc_skb".

     with, UDP traffic no kernel panic at all. I am using redhat 6.2 and the
kernel 2.4.1.

any help greatly appreciated,
regards,
ntkarthik

-----------------------------------------------------------------------        
                Karthikeyan Nathillvar


Office:                                                               Res.:
-------                                                         -----
Team Niehaus Graduate Research Assistant     
245 E, Nichols Hall,                                   1550, Eddingham Drive,
2335 Irving Hill Road,                                 Lawrence,
Lawrence,                                                 Kansas 66046 -4037
Kansas 66045 - 7612                                 Ph: 785-312-8001
Ph: 785-864-7774
               
                         " Dare to Dream Care to Achieve"
-------------------------------------------------------------------------------

Re: [mpls-linux-general] MPLS and ATM

[James R. L. <jl...@mi...>]

Hello,

What is the atm index of your ATM card?

cat /proc/net/atm/device

It shows up as the Intf.

When creating your atm interface (via atmarp -c) the index in the name
needs to match that of the index from above.

Example:

If the atm card index (from /proc/net/atm/device) is 3

atmarp -c atm3

I know this is a kind of a ridiculous rule, but due to the nature of
ATM for Linux it is a requirement.  Someday I'll get a around to re-writing
the way ATM interfaces are handled on Linux ...

Let me know if this still doesn't help.

Jim

On Wed, Apr 04, 2001 at 05:51:21PM -0400, Jonathan Earle wrote:
> Hi James,
> 
> Setup is as follows:
> 
> Interphase 5575 ATM NIC
> Kernel 2.4.3
> MPLS v0.990
> 
> Basic IP connectivity between the two NICs works fine.  I configure the NICs
> thusly:
> 
> master)
>        atmsigd -b -m network
>        atmarpd -b
>        sleep 2s
> 
>        if [ "$adapter" != "atm$anum" ]; then
>           atmarp -c atm$anum
>        fi
> 
>        ifconfig atm$anum 192.168.1.1 up
>        sleep 2s
>        atmarp -s 192.168.1.2 0.0.102
>        ;;
>  slave)
>        atmsigd -b
>        atmarpd -b
>        sleep 2s
> 
>        if [ "$adapter" != "atm$anum" ]; then
>           atmarp -c atm$anum
>        fi
> 
>        ifconfig atm$anum 192.168.1.2 up
>        sleep 2s
>        atmarp -s 192.168.1.1 0.0.102
>        ;;
> 
> Using the examples in REAME.atm, I can create the labelspace:
> 
> root@onc4:~> cat /proc/net/mpls_labelspace
> lo      0
> teql0   0
> atm1    3
> 
> But not IN labels:
> 
> mpls_add_in_label: enter
> mpls_info_default_in_instruction: enter
> mpls_info_default_in_instruction: exit
> mpls_atm_create_vcc: enter(1,0,33,1)
> mpls_atm_create_vcc: adev == NULL
> mpls_add_in_label: failed creating vcc
> __mpls_del_in_label: enter
> __mpls_del_in_label: exit
> mpls_add_in_label: exit
> 
> or OUT labels:
> 
> mpls_add_out_label: enter
> mpls_info_default_moi_instruction: enter
> mpls_info_default_moi_instruction: exit
> mpls_atm_create_vcc: enter(1,0,33,2)
> mpls_atm_create_vcc: adev == NULL
> mpls_add_out_label: failed to create VCC
> __mpls_del_out_label: enter
> __mpls_del_out_label: exit
> mpls_add_out_label: exit
> 
> Cheers!
> Jon
> 
> _______________________________________________
> mpls-linux-general mailing list
> mpl...@li...
> http://lists.sourceforge.net/lists/listinfo/mpls-linux-general

-- 
James R. Leu

Re: [mpls-linux-general] MPLS on ATM

[James R. L. <jl...@mi...>]

Hello,

Great output.  It lets me know exactly what is happening.
It looks like your creating a good outlabel.  The inlabel looks like it is
created as well. Packets are being sent from MPLS to the ATM layer on the
ingress.  But on the egress, nothing is ever making it to the MPLS layer.

Can you get cell counters on your ATM switch?  Check to see if it is seeing
any cells coming from the ingress.  Also can you check the ATM counters
on the egress and see if any cells are making it that far.

Can you try doing a back to back connection? (no switch)

Jim

On Wed, Apr 04, 2001 at 04:22:55PM +0100, cacote wrote:
> HI,
> 
> The message debug output of the two hosts ( NORTE e ESTE ) used in the
> scenario described, is included in the two files that go in attach.
> 
> The file nortemes is the the message of the Host  Norte (machine that does
> the pinging ) and the file estemes
> is the message output of the Host Este ( machine that is pinged  and that
> crashes ).
> 
> I dont know if it helps you in any way but, we have the two ATM  cards in
> the scenario connected by a switch in each we created the PVC  we are using (
> 0/104 PVC).
> 
>                                                 Thank you for your help ,
> best regards
> 
> 
> HUGO CACOTE
> 
> 
> 
> "James R. Leu" wrote:
> 
> > BTW
> >
> > Do you have any output from the crashes?  I'd like to find where it is
> > crashing, and fix it :-)
> >
> > Make sure to send me the MPLS debug output.
> >
> > Jim
> >
> >   On Tue, Apr 03, 2001 at 08:38:58PM +0100, cacote wrote:
> > > HI
> > >
> > > I ve been testing the new MPLS package (mpls-linux-0.990), it worked
> > > allright in ethernet but when I
> > > got to teste it in ATM there were problems.
> > >
> > > My scenario is the following:
> > >
> > >
> > >
> > > |||||||||||||||||
> > > ||||||||||||||||
> > >   ||  norte
> > > ||
> > > ||   este  ||
> > >   ||
> > > ||
> > > ||            ||
> > >
> > > |||||||||||||||||
> > > ||||||||||||||||
> > >          |   atm0
> > > 192.168.50.10                                                    |  atm0
> > > 192.168.50.20
> > >
> > > |
> > > |
> > >          |------------------------------------- |
> > >
> > > The ATM  cards I am using are the following:
> > >
> > > ForeRunnerLe 155 mbit/s          nicstar  (chipset)
> > >
> > > The configuration  i made on both hosts  was this:
> > >
> > >
> > > ON norte:
> > > [root@norte /root]# route add -host 192.168.50.20/32 gw 192.168.50.20
> > > [root@norte /root]# mplsadm -A -O atm:0/104:atm0
> > > [root@norte /root]# mplsadm -B -O atm:0/104:atm0 -f 192.168.50.20/32
> > > [root@norte /root]# mplsadm -L atm0:0
> > >
> > > If we try and check the configurations we see:
> > >
> > > [root@norte /root]# cat /proc/net/mpls_out
> > > 80680008 PUSH(atm 0/104) SET(atm0)
> > > [root@norte /root]# cat /proc/net/mpls_fec
> > > 80680008 192.168.50.20/32
> > >
> > > ON este:
> > > [root@este /root]# route add -host 192.168.50.10/32 gw 192.168.50.10
> > > [root@este /root]# mplsadm -L atm0:0
> > > [root@este /root]# mplsadm -A -I atm:0/104:0
> > >
> > > If we try and check the configurations we see:
> > >
> > > [root@este /root]# cat /proc/net/mpls_in
> > > 80680000 atm 0/104 0 POP DLV
> > >
> > > If we then make  [root@norte /root]# ping 192.168.50.20 , and if then
> > > use  ethereal to monitor the
> > > packets in the two hosts we see no MPLS and just ICMP. I should add that
> > > sometimes whem trying this
> > > scenario the machine that is being pinged just crashes with no apparent
> > > reason.
> > >
> > > Is there something wrong in the setup of the scenario? Is there anybody
> > > that has a scenario where MPLS
> > > works with ATM ?
> > >
> > >     Thank you for the Help, best regards
> > >
> > >             HUGO CACOTE
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > > _______________________________________________
> > > mpls-linux-general mailing list
> > > mpl...@li...
> > > http://lists.sourceforge.net/lists/listinfo/mpls-linux-general
> >
> > --
> > James R. Leu

> Apr  4 15:14:18 este kernel: mpls_set_labelspace: enter
> Apr  4 15:14:18 este kernel: mpls_set_labelspace: labelspace(0)
> Apr  4 15:14:18 este kernel: mpls_set_labelspace: exit
> Apr  4 15:16:26 este kernel: mpls_add_in_label: enter
> Apr  4 15:16:26 este kernel: mpls_info_default_in_instruction: enter
> Apr  4 15:16:26 este kernel: mpls_info_default_in_instruction: exit
> Apr  4 15:16:26 este kernel: mpls_atm_create_vcc: enter(0,0,104,1)
> Apr  4 15:16:26 este kernel: mpls_atm_create_vcc: vci 104 <-> 1024 vpi 0 <-> 4
> Apr  4 15:16:26 este kernel: mpls_atm_create_vcc: create_vcc
> Apr  4 15:16:26 este kernel: mpls_atm_create_vcc: setting up vcc
> Apr  4 15:16:26 este kernel: mpls_atm_create_vcc: exit
> Apr  4 15:16:26 este kernel: mpls_add_in_label: succeeded creating vcc
> Apr  4 15:16:26 este kernel: mpls_add_in_label: exit

> Apr  4 15:21:23 norte kernel: mpls_set_labelspace: enter
> Apr  4 15:21:23 norte kernel: mpls_set_labelspace: labelspace(0)
> Apr  4 15:21:23 norte kernel: mpls_set_labelspace: exit
> Apr  4 15:24:53 norte kernel: mpls_add_out_label: enter
> Apr  4 15:24:53 norte kernel: mpls_make_dst: enter
> Apr  4 15:24:53 norte kernel: mpls_make_dst: before bind neighbor
> Apr  4 15:24:53 norte kernel: mpls_make_dst: bind neighbor succeeded
> Apr  4 15:24:53 norte kernel: mpls_make_dst: exit(c3e2a380)
> Apr  4 15:24:53 norte kernel: mpls_info_default_moi_instruction: enter
> Apr  4 15:24:53 norte kernel: mpls_info_default_moi_instruction: exit
> Apr  4 15:24:53 norte kernel: mpls_atm_create_vcc: enter(0,0,104,2)
> Apr  4 15:24:53 norte kernel: mpls_atm_create_vcc: vci 104 <-> 1024 vpi 0 <-> 4
> Apr  4 15:24:53 norte kernel: mpls_atm_create_vcc: create_vcc
> Apr  4 15:24:53 norte kernel: mpls_atm_create_vcc: setting up vcc
> Apr  4 15:24:53 norte kernel: mpls_atm_create_vcc: exit
> Apr  4 15:24:53 norte kernel: mpls_add_out_label: succeeded to create VCC
> Apr  4 15:24:53 norte kernel: Label ATM 0/104
> Apr  4 15:24:53 norte kernel: Key ATM 0/104 8
> Apr  4 15:24:53 norte kernel: mpls_add_out_label: exit
> Apr  4 15:24:53 norte kernel: mpls_prep_out2fec: enter
> Apr  4 15:24:53 norte kernel: Label ATM 0/104
> Apr  4 15:24:53 norte kernel: Key ATM 0/104 8
> Apr  4 15:24:53 norte kernel: mpls_fill_key_res: enter
> Apr  4 15:24:53 norte kernel: mpls_fill_key_res: exit
> Apr  4 15:24:53 norte kernel: mpls_bind_out2fec: copied label to the FIB_NODE
> Apr  4 15:24:53 norte kernel: mpls_bind_out2fec: Added FEC to list
> Apr  4 15:24:53 norte kernel: mpls_bind_out2fec: exit
> Apr  4 15:26:01 norte kernel: rt_set_nexthop: adding label to dst
> Apr  4 15:26:01 norte kernel: rt_set_nexthop: before - mtu = 9180
> Apr  4 15:26:01 norte kernel: rt_set_nexthop: after - mtu = 9176
> Apr  4 15:26:01 norte kernel: rt_set_nexthop: adding label to dst
> Apr  4 15:26:01 norte kernel: rt_set_nexthop: before - mtu = 9180
> Apr  4 15:26:01 norte kernel: rt_set_nexthop: after - mtu = 9176
> Apr  4 15:26:02 norte kernel: mpls_output: enter
> Apr  4 15:26:02 norte kernel: mpls_output2: enter
> Apr  4 15:26:02 norte kernel: mpls_output2: push
> Apr  4 15:26:02 norte kernel: mpls_opcode_push: enter
> Apr  4 15:26:02 norte kernel: mpls_opcode_push: using tailroom
> Apr  4 15:26:02 norte kernel: mpls_opcode_push: done using tailroom
> Apr  4 15:26:02 norte kernel: mpls_opcode_push: exit
> Apr  4 15:26:02 norte kernel: mpls_output2: set ATM
> Apr  4 15:26:02 norte kernel: mpls_output2: output device = atm0
> Apr  4 15:26:02 norte kernel: mpls_finish: enter
> Apr  4 15:26:02 norte kernel: mpls_finish: exit
> Apr  4 15:26:02 norte kernel: mpls_output2: ATM
> Apr  4 15:26:02 norte kernel: mpls_output2: ATM send
> Apr  4 15:26:02 norte kernel: mpls_output2: exit(0)
> Apr  4 15:26:02 norte kernel: mpls_atm_pop: enter
> Apr  4 15:26:02 norte kernel: mpls_atm_pop: exit
> Apr  4 15:26:02 norte kernel: mpls_output: exit
> Apr  4 15:26:03 norte kernel: mpls_output: enter
> Apr  4 15:26:03 norte kernel: mpls_output2: enter
> Apr  4 15:26:03 norte kernel: mpls_output2: push
> Apr  4 15:26:03 norte kernel: mpls_opcode_push: enter
> Apr  4 15:26:03 norte kernel: mpls_opcode_push: using tailroom
> Apr  4 15:26:03 norte kernel: mpls_opcode_push: done using tailroom
> Apr  4 15:26:03 norte kernel: mpls_opcode_push: exit
> Apr  4 15:26:03 norte kernel: mpls_output2: set ATM
> Apr  4 15:26:03 norte kernel: mpls_output2: output device = atm0
> Apr  4 15:26:03 norte kernel: mpls_finish: enter
> Apr  4 15:26:03 norte kernel: mpls_finish: exit
> Apr  4 15:26:03 norte kernel: mpls_output2: ATM
> Apr  4 15:26:03 norte kernel: mpls_output2: ATM send
> Apr  4 15:26:03 norte kernel: mpls_output2: exit(0)
> Apr  4 15:26:03 norte kernel: mpls_atm_pop: enter
> Apr  4 15:26:03 norte kernel: mpls_atm_pop: exit
> Apr  4 15:26:03 norte kernel: mpls_output: exit
> Apr  4 15:26:04 norte kernel: mpls_output: enter
> Apr  4 15:26:04 norte kernel: mpls_output2: enter
> Apr  4 15:26:04 norte kernel: mpls_output2: push
> Apr  4 15:26:04 norte kernel: mpls_opcode_push: enter
> Apr  4 15:26:04 norte kernel: mpls_opcode_push: using tailroom
> Apr  4 15:26:04 norte kernel: mpls_opcode_push: done using tailroom
> Apr  4 15:26:04 norte kernel: mpls_opcode_push: exit
> Apr  4 15:26:04 norte kernel: mpls_output2: set ATM
> Apr  4 15:26:04 norte kernel: mpls_output2: output device = atm0
> Apr  4 15:26:04 norte kernel: mpls_finish: enter
> Apr  4 15:26:04 norte kernel: mpls_finish: exit
> Apr  4 15:26:04 norte kernel: mpls_output2: ATM
> Apr  4 15:26:04 norte kernel: mpls_output2: ATM send
> Apr  4 15:26:04 norte kernel: mpls_output2: exit(0)
> Apr  4 15:26:04 norte kernel: mpls_atm_pop: enter
> Apr  4 15:26:04 norte kernel: mpls_atm_pop: exit
> Apr  4 15:26:04 norte kernel: mpls_output: exit
> Apr  4 15:26:05 norte kernel: mpls_output: enter
> Apr  4 15:26:05 norte kernel: mpls_output2: enter
> Apr  4 15:26:05 norte kernel: mpls_output2: push
> Apr  4 15:26:05 norte kernel: mpls_opcode_push: enter
> Apr  4 15:26:05 norte kernel: mpls_opcode_push: using tailroom
> Apr  4 15:26:05 norte kernel: mpls_opcode_push: done using tailroom


-- 
James R. Leu

[mpls-linux-general] MPLS and ATM

[Jonathan E. <je...@no...>]

Hi James,

Setup is as follows:

Interphase 5575 ATM NIC
Kernel 2.4.3
MPLS v0.990

Basic IP connectivity between the two NICs works fine.  I configure the NICs
thusly:

master)
       atmsigd -b -m network
       atmarpd -b
       sleep 2s

       if [ "$adapter" != "atm$anum" ]; then
          atmarp -c atm$anum
       fi

       ifconfig atm$anum 192.168.1.1 up
       sleep 2s
       atmarp -s 192.168.1.2 0.0.102
       ;;
 slave)
       atmsigd -b
       atmarpd -b
       sleep 2s

       if [ "$adapter" != "atm$anum" ]; then
          atmarp -c atm$anum
       fi

       ifconfig atm$anum 192.168.1.2 up
       sleep 2s
       atmarp -s 192.168.1.1 0.0.102
       ;;

Using the examples in REAME.atm, I can create the labelspace:

root@onc4:~> cat /proc/net/mpls_labelspace
lo      0
teql0   0
atm1    3

But not IN labels:

mpls_add_in_label: enter
mpls_info_default_in_instruction: enter
mpls_info_default_in_instruction: exit
mpls_atm_create_vcc: enter(1,0,33,1)
mpls_atm_create_vcc: adev == NULL
mpls_add_in_label: failed creating vcc
__mpls_del_in_label: enter
__mpls_del_in_label: exit
mpls_add_in_label: exit

or OUT labels:

mpls_add_out_label: enter
mpls_info_default_moi_instruction: enter
mpls_info_default_moi_instruction: exit
mpls_atm_create_vcc: enter(1,0,33,2)
mpls_atm_create_vcc: adev == NULL
mpls_add_out_label: failed to create VCC
__mpls_del_out_label: enter
__mpls_del_out_label: exit
mpls_add_out_label: exit

Cheers!
Jon