[support] UMIP handover problem

[Gábor Hegedüs]

Dear All!

I use UMIP MIPv6 testben in a virtual network. I start the MN in a foreign
network.
It is working well. Ping and traceroute are OK.
But when I try to make a handover the HA say: tunnel deleted. I can see 2-3
icmp message before the tunnel deleted message.

When I make the handover I do not change the attachment point of the MN. I
just delete the linc-local and global address of the router and kill radvd.
After that I add a new link-local and a global address to the interface and
I run radvd with a new config file.

Topology:

<HA>------<router>---[MN]

Can anybody give me an advice to solve this problem?

Thanks,
G?bor
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://ml.nautilus6.org/pipermail/support/attachments/20131128/15adc3ff/attachment.html>

[Romain KUNTZ]

Hello,

Can you send the logs of mip6d from the HA and the MN? This will help understanding where the problem comes from.

Thank you,
Romain

Post by Gábor Hegedüs
Dear All!
I use UMIP MIPv6 testben in a virtual network. I start the MN in a foreign network.
It is working well. Ping and traceroute are OK.
But when I try to make a handover the HA say: tunnel deleted. I can see 2-3 icmp message before the tunnel deleted message.
When I make the handover I do not change the attachment point of the MN. I just delete the linc-local and global address of the router and kill radvd. After that I add a new link-local and a global address to the interface and I run radvd with a new config file.
<HA>------<router>---[MN]
Can anybody give me an advice to solve this problem?
Thanks,
G?bor
_______________________________________________
Support mailing list
Support at ml.nautilus6.org
http://ml.nautilus6.org/mailman/listinfo/support

[Romain KUNTZ]

Hello Gabor,

(Please keep the mailing list in the loop).

The problems seems to come from a routing issue.

It seems that the first registration is successful (with CoA 2001:baba:2001:bab2:a00:27ff:fedc:ba1) then the registration update in the new network is also OK (with CoA 2001:baba:2001:bab4:a00:27ff:fedc:ba1), but after that, the MN keeps sending a refresh BU that never reaches the HA: on the MN log side you can see the following messages:

Thu Nov 28 15:47:15 bu_refresh: BU refresh type: 0
[?]
Thu Nov 28 15:47:16 bu_resend: BU resend type 0
[?]
Thu Nov 28 15:47:18 bu_resend: BU resend type 0
[?]

which indicates that no BA was received. On the HA side, you can also see that the HA never receives any refresh BU, and thus deletes the tunnel because registration has reached the timeout.

Can you check on your router if routing is correctly performed over time? It seems that your refresh BU are lost somewhere between the MR and the HA.

Romain

Hi Romain!
I attached the logs. I do not use IPsec in this logs but I enabled IPsec once yesterday and I realized something strange. I saw correct ncoa and ocoa, it was OK. But before the tunnel deleted message I saw ncoa is equal the home address of the MN's address but teh MN was not in the home network. Maybe it can be the problem. I did not save the IPsec logs. If you want I save and sand you those logs too.
Thanks,
G?bor
2013/11/29 Romain KUNTZ <r.kuntz at gmail.com>
Hello,
Can you send the logs of mip6d from the HA and the MN? This will help understanding where the problem comes from.
Thank you,
Romain

Post by Gábor Hegedüs
Dear All!
I use UMIP MIPv6 testben in a virtual network. I start the MN in a foreign network.
It is working well. Ping and traceroute are OK.
But when I try to make a handover the HA say: tunnel deleted. I can see 2-3 icmp message before the tunnel deleted message.
When I make the handover I do not change the attachment point of the MN. I just delete the linc-local and global address of the router and kill radvd. After that I add a new link-local and a global address to the interface and I run radvd with a new config file.
<HA>------<router>---[MN]
Can anybody give me an advice to solve this problem?
Thanks,
G?bor
_______________________________________________
Support mailing list
Support at ml.nautilus6.org
http://ml.nautilus6.org/mailman/listinfo/support

<HAlog.txt><MNlog.txt>

[Bokor László]

Dear Gabor,

based on Romain's comments and your testbed setup I believe that the new
prefix (advertised with the changed RA config after the such emulated
handover) is not routed appropriately. If I am right, you just should
prepare the routes also for the new prefix.

Best regards,
goodzi



2013/11/29 Romain KUNTZ <r.kuntz at gmail.com>

Post by Romain KUNTZ
Hello Gabor,
(Please keep the mailing list in the loop).
The problems seems to come from a routing issue.
It seems that the first registration is successful (with CoA
2001:baba:2001:bab2:a00:27ff:fedc:ba1) then the registration update in the
new network is also OK (with CoA 2001:baba:2001:bab4:a00:27ff:fedc:ba1),
but after that, the MN keeps sending a refresh BU that never reaches the
Thu Nov 28 15:47:15 bu_refresh: BU refresh type: 0
[?]
Thu Nov 28 15:47:16 bu_resend: BU resend type 0
[?]
Thu Nov 28 15:47:18 bu_resend: BU resend type 0
[?]
which indicates that no BA was received. On the HA side, you can also see
that the HA never receives any refresh BU, and thus deletes the tunnel
because registration has reached the timeout.
Can you check on your router if routing is correctly performed over time?
It seems that your refresh BU are lost somewhere between the MR and the HA.
Romain

Hi Romain!
I attached the logs. I do not use IPsec in this logs but I enabled IPsec

once yesterday and I realized something strange. I saw correct ncoa and
ocoa, it was OK. But before the tunnel deleted message I saw ncoa is equal
the home address of the MN's address but teh MN was not in the home
network. Maybe it can be the problem. I did not save the IPsec logs. If you
want I save and sand you those logs too.

Thanks,
G?bor
2013/11/29 Romain KUNTZ <r.kuntz at gmail.com>
Hello,
Can you send the logs of mip6d from the HA and the MN? This will help

understanding where the problem comes from.

Thank you,
Romain

Post by Gábor Hegedüs
Dear All!
I use UMIP MIPv6 testben in a virtual network. I start the MN in a

foreign network.

Post by Gábor Hegedüs
It is working well. Ping and traceroute are OK.
But when I try to make a handover the HA say: tunnel deleted. I can

see 2-3 icmp message before the tunnel deleted message.

Post by Gábor Hegedüs
When I make the handover I do not change the attachment point of the

MN. I just delete the linc-local and global address of the router and kill
radvd. After that I add a new link-local and a global address to the
interface and I run radvd with a new config file.

Post by Gábor Hegedüs
<HA>------<router>---[MN]
Can anybody give me an advice to solve this problem?
Thanks,
G?bor
_______________________________________________
Support mailing list
Support at ml.nautilus6.org
http://ml.nautilus6.org/mailman/listinfo/support

<HAlog.txt><MNlog.txt>

_______________________________________________
Support mailing list
Support at ml.nautilus6.org
http://ml.nautilus6.org/mailman/listinfo/support

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://ml.nautilus6.org/pipermail/support/attachments/20131130/d136b0e0/attachment.html>

[Gábor Hegedüs]

HI!

Problem solved.
I used global addresses to create routes between nodes.
Now I am using link-local addresses and default routet from tahe touter to
HA and
from HA to foreign networks.
And I setted up this lines correctly:

net.ipv6.conf.default.autoconf
net.ipv6.conf.default.accept_ra
net.ipv6.conf.all.autoconf
net.ipv6.conf.all.accept_ra
net.ipv6.conf.eth0.autoconf
net.ipv6.conf.eth0.accept_ra

Thanks for everyone,
G?bor


2013/12/5 G?bor Heged?s <hgabor1989 at gmail.com>

HI!
Problem solved.
I used global addresses to create routes between nodes.
Now I am using link-local addresses and default routet from tahe touter to
HA and
from HA to foreign networks.
net.ipv6.conf.default.autoconf
net.ipv6.conf.default.accept_ra
net.ipv6.conf.all.autoconf
net.ipv6.conf.all.accept_ra
net.ipv6.conf.eth0.autoconf
net.ipv6.conf.eth0.accept_ra
Thanks for everyone,
G?bor
2013/11/30 G?bor Heged?s <hgabor1989 at gmail.com>

Hello!
I think the routing is ok. When I ping the HA from MN it works with the
original prefix.
When I ping the HA after the new prefix advertised I can't reach the HA
or any other address outside.
But when I turn off and on the interface the ping works again (with the
new prefix).
When the new RA comes the MN has 2-3 global addresses. one or two from
the old RA and one from the new RA.
Is it problem? If it is a problrm how could I solve it? (I kill the old
radvd before start of teh new radvd).
Thanks,
G?bor
2013/11/30 Bokor L?szl? <goodzi at gmail.com>

Post by Bokor László
Dear Gabor,
based on Romain's comments and your testbed setup I believe that the new
prefix (advertised with the changed RA config after the such emulated
handover) is not routed appropriately. If I am right, you just should
prepare the routes also for the new prefix.
Best regards,
goodzi
2013/11/29 Romain KUNTZ <r.kuntz at gmail.com>

Post by Romain KUNTZ
Hello Gabor,
(Please keep the mailing list in the loop).
The problems seems to come from a routing issue.
It seems that the first registration is successful (with CoA
2001:baba:2001:bab2:a00:27ff:fedc:ba1) then the registration update in the
new network is also OK (with CoA 2001:baba:2001:bab4:a00:27ff:fedc:ba1),
but after that, the MN keeps sending a refresh BU that never reaches the
Thu Nov 28 15:47:15 bu_refresh: BU refresh type: 0
[?]
Thu Nov 28 15:47:16 bu_resend: BU resend type 0
[?]
Thu Nov 28 15:47:18 bu_resend: BU resend type 0
[?]
which indicates that no BA was received. On the HA side, you can also
see that the HA never receives any refresh BU, and thus deletes the tunnel
because registration has reached the timeout.
Can you check on your router if routing is correctly performed over
time? It seems that your refresh BU are lost somewhere between the MR and
the HA.
Romain

Hi Romain!
I attached the logs. I do not use IPsec in this logs but I enabled

IPsec once yesterday and I realized something strange. I saw correct ncoa
and ocoa, it was OK. But before the tunnel deleted message I saw ncoa is
equal the home address of the MN's address but teh MN was not in the home
network. Maybe it can be the problem. I did not save the IPsec logs. If you
want I save and sand you those logs too.

Thanks,
G?bor
2013/11/29 Romain KUNTZ <r.kuntz at gmail.com>
Hello,
Can you send the logs of mip6d from the HA and the MN? This will help

understanding where the problem comes from.

Thank you,
Romain
On 28 Nov 2013, at 16:19 , G?bor Heged?s <hgabor1989 at gmail.com>

Post by Gábor Hegedüs
Dear All!
I use UMIP MIPv6 testben in a virtual network. I start the MN in a

foreign network.

Post by Gábor Hegedüs
It is working well. Ping and traceroute are OK.
But when I try to make a handover the HA say: tunnel deleted. I can

see 2-3 icmp message before the tunnel deleted message.

Post by Gábor Hegedüs
When I make the handover I do not change the attachment point of

the MN. I just delete the linc-local and global address of the router and
kill radvd. After that I add a new link-local and a global address to the
interface and I run radvd with a new config file.

Post by Gábor Hegedüs
<HA>------<router>---[MN]
Can anybody give me an advice to solve this problem?
Thanks,
G?bor
_______________________________________________
Support mailing list
Support at ml.nautilus6.org
http://ml.nautilus6.org/mailman/listinfo/support

<HAlog.txt><MNlog.txt>

_______________________________________________
Support mailing list
Support at ml.nautilus6.org
http://ml.nautilus6.org/mailman/listinfo/support

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://ml.nautilus6.org/pipermail/support/attachments/20131205/cae136a7/attachment.html>

[Romain KUNTZ]

Hi Gabor,

Thanks for telling us, I?m glad yo could fix your issues! Don?t hesitate to post on the mailing list in the future if need be.

Best,
Romain

Post by Gábor Hegedüs
HI!
Problem solved.
I used global addresses to create routes between nodes.
Now I am using link-local addresses and default routet from tahe touter to HA and
from HA to foreign networks.
net.ipv6.conf.default.autoconf
net.ipv6.conf.default.accept_ra
net.ipv6.conf.all.autoconf
net.ipv6.conf.all.accept_ra
net.ipv6.conf.eth0.autoconf
net.ipv6.conf.eth0.accept_ra
Thanks for everyone,
G?bor
2013/12/5 G?bor Heged?s <hgabor1989 at gmail.com>
HI!
Problem solved.
I used global addresses to create routes between nodes.
Now I am using link-local addresses and default routet from tahe touter to HA and
from HA to foreign networks.
net.ipv6.conf.default.autoconf
net.ipv6.conf.default.accept_ra
net.ipv6.conf.all.autoconf
net.ipv6.conf.all.accept_ra
net.ipv6.conf.eth0.autoconf
net.ipv6.conf.eth0.accept_ra
Thanks for everyone,
G?bor
2013/11/30 G?bor Heged?s <hgabor1989 at gmail.com>
Hello!
I think the routing is ok. When I ping the HA from MN it works with the original prefix.
When I ping the HA after the new prefix advertised I can't reach the HA or any other address outside.
But when I turn off and on the interface the ping works again (with the new prefix).
When the new RA comes the MN has 2-3 global addresses. one or two from the old RA and one from the new RA.
Is it problem? If it is a problrm how could I solve it? (I kill the old radvd before start of teh new radvd).
Thanks,
G?bor
2013/11/30 Bokor L?szl? <goodzi at gmail.com>
Dear Gabor,
based on Romain's comments and your testbed setup I believe that the new prefix (advertised with the changed RA config after the such emulated handover) is not routed appropriately. If I am right, you just should prepare the routes also for the new prefix.
Best regards,
goodzi
2013/11/29 Romain KUNTZ <r.kuntz at gmail.com>
Hello Gabor,
(Please keep the mailing list in the loop).
The problems seems to come from a routing issue.
Thu Nov 28 15:47:15 bu_refresh: BU refresh type: 0
[?]
Thu Nov 28 15:47:16 bu_resend: BU resend type 0
[?]
Thu Nov 28 15:47:18 bu_resend: BU resend type 0
[?]
which indicates that no BA was received. On the HA side, you can also see that the HA never receives any refresh BU, and thus deletes the tunnel because registration has reached the timeout.
Can you check on your router if routing is correctly performed over time? It seems that your refresh BU are lost somewhere between the MR and the HA.
Romain

Hi Romain!
I attached the logs. I do not use IPsec in this logs but I enabled IPsec once yesterday and I realized something strange. I saw correct ncoa and ocoa, it was OK. But before the tunnel deleted message I saw ncoa is equal the home address of the MN's address but teh MN was not in the home network. Maybe it can be the problem. I did not save the IPsec logs. If you want I save and sand you those logs too.
Thanks,
G?bor
2013/11/29 Romain KUNTZ <r.kuntz at gmail.com>
Hello,
Can you send the logs of mip6d from the HA and the MN? This will help understanding where the problem comes from.
Thank you,
Romain

Post by Gábor Hegedüs
Dear All!
I use UMIP MIPv6 testben in a virtual network. I start the MN in a foreign network.
It is working well. Ping and traceroute are OK.
But when I try to make a handover the HA say: tunnel deleted. I can see 2-3 icmp message before the tunnel deleted message.
When I make the handover I do not change the attachment point of the MN. I just delete the linc-local and global address of the router and kill radvd. After that I add a new link-local and a global address to the interface and I run radvd with a new config file.
<HA>------<router>---[MN]
Can anybody give me an advice to solve this problem?
Thanks,
G?bor
_______________________________________________
Support mailing list
Support at ml.nautilus6.org
http://ml.nautilus6.org/mailman/listinfo/support

<HAlog.txt><MNlog.txt>

_______________________________________________
Support mailing list
Support at ml.nautilus6.org
http://ml.nautilus6.org/mailman/listinfo/support
_______________________________________________
Support mailing list
Support at ml.nautilus6.org
http://ml.nautilus6.org/mailman/listinfo/support