Mailing List Archive

IPv4 BFD flaps on MX204
Hi,

I have a MX204 with a couple of OSPF/OSPFv3 adj to other MXs over two
et- interfaces, BFD enabled for both protocols.
At random intervals the IPv4 BFD sessions are flapping with or without
the 'no-delegate-processing' option, however the IPv6 sessions are
stable (they are using the same timers).

Did anyone experienced something like this? The router is running 18.2R3-S3.


Thanks!
_______________________________________________
juniper-nsp mailing list juniper-nsp@puck.nether.net
https://puck.nether.net/mailman/listinfo/juniper-nsp
Re: IPv4 BFD flaps on MX204 [ In reply to ]
On 18/Aug/20 01:35, Mihai wrote:

> Hi,
>
> I have a MX204 with a couple of OSPF/OSPFv3 adj to other MXs over two
> et- interfaces, BFD enabled for both protocols.
> At random intervals the IPv4 BFD sessions are flapping with or without
> the 'no-delegate-processing' option, however the IPv6 sessions are
> stable (they are using the same timers).
>
> Did anyone experienced something like this? The router is running
> 18.2R3-S3.

Do you have BFDv6 enabled as well?

Mark.
_______________________________________________
juniper-nsp mailing list juniper-nsp@puck.nether.net
https://puck.nether.net/mailman/listinfo/juniper-nsp
Re: IPv4 BFD flaps on MX204 [ In reply to ]
If you have many uncontrolled directly connected L2 domains like IX-es or customers, check policer __default_arp_policer__.
If there are drops, you need to apply interface-specific arp policers to interfaces with protocols and/or to source of arp bursts.



18.08.2020 02:35, Mihai ?????:
> Hi,
>
> I have a MX204 with a couple of OSPF/OSPFv3 adj to other MXs over two et- interfaces, BFD enabled for both protocols.
> At random intervals the IPv4 BFD sessions are flapping with or without the 'no-delegate-processing' option, however the IPv6 sessions are stable (they are using the same timers).
>
> Did anyone experienced something like this? The router is running 18.2R3-S3.
>
>
> Thanks!
> _______________________________________________
> juniper-nsp mailing list juniper-nsp@puck.nether.net
> https://puck.nether.net/mailman/listinfo/juniper-nsp

--

_______________________________________________
juniper-nsp mailing list juniper-nsp@puck.nether.net
https://puck.nether.net/mailman/listinfo/juniper-nsp
Re: IPv4 BFD flaps on MX204 [ In reply to ]
Hi Ivan,

Indeed, that was the issue, thanks!

Regards

On 18/08/2020 09:54, Ivan Malyarchuk wrote:
> If you have many uncontrolled directly connected L2 domains like IX-es
> or customers, check policer __default_arp_policer__.
> If there are drops, you need to apply interface-specific arp policers to
> interfaces with protocols and/or to source of arp bursts.
>
>
>
> 18.08.2020 02:35, Mihai ?????:
>> Hi,
>>
>> I have a MX204 with a couple of OSPF/OSPFv3 adj to other MXs over two
>> et- interfaces, BFD enabled for both protocols.
>> At random intervals the IPv4 BFD sessions are flapping with or without
>> the 'no-delegate-processing' option, however the IPv6 sessions are
>> stable (they are using the same timers).
>>
>> Did anyone experienced something like this? The router is running
>> 18.2R3-S3.
>>
>>
>> Thanks!
>> _______________________________________________
>> juniper-nsp mailing list juniper-nsp@puck.nether.net
>> https://puck.nether.net/mailman/listinfo/juniper-nsp
>

_______________________________________________
juniper-nsp mailing list juniper-nsp@puck.nether.net
https://puck.nether.net/mailman/listinfo/juniper-nsp
Re: IPv4 BFD flaps on MX204 [ In reply to ]
I'd recommend disabling IPv6 BFD for now as well. It's not handled in
hardware, and so if it drops, you would lose your IGP. Worse if you run
multiple address families in your IGP, e.g., IS-IS, or IPv4 over OSPFv3 .

Hardware support for BFDv6 is coming in 1H'21.

Mark.

On 18/Aug/20 13:01, Mihai wrote:
> Hi Ivan,
>
> Indeed, that was the issue, thanks!
>
> Regards
>
> On 18/08/2020 09:54, Ivan Malyarchuk wrote:
>> If you have many uncontrolled directly connected L2 domains like
>> IX-es or customers, check policer __default_arp_policer__.
>> If there are drops, you need to apply interface-specific arp policers
>> to interfaces with protocols and/or to source of arp bursts.
>>
>>
>>
>> 18.08.2020 02:35, Mihai ?????:
>>> Hi,
>>>
>>> I have a MX204 with a couple of OSPF/OSPFv3 adj to other MXs over
>>> two et- interfaces, BFD enabled for both protocols.
>>> At random intervals the IPv4 BFD sessions are flapping with or
>>> without the 'no-delegate-processing' option, however the IPv6
>>> sessions are stable (they are using the same timers).
>>>
>>> Did anyone experienced something like this? The router is running
>>> 18.2R3-S3.
>>>
>>>
>>> Thanks!
>>> _______________________________________________
>>> juniper-nsp mailing list juniper-nsp@puck.nether.net
>>> https://puck.nether.net/mailman/listinfo/juniper-nsp
>>
>
> _______________________________________________
> juniper-nsp mailing list juniper-nsp@puck.nether.net
> https://puck.nether.net/mailman/listinfo/juniper-nsp

_______________________________________________
juniper-nsp mailing list juniper-nsp@puck.nether.net
https://puck.nether.net/mailman/listinfo/juniper-nsp
Re: IPv4 BFD flaps on MX204 [ In reply to ]
Hi,

In fact inline IPv6 BFD is supported for other than link-local addresses starting from 18.1 IIRC. This doesn't help for IS-IS or OSPFv3, though, as those use link-local addresses for adjacencies.

We do have IPv6 BFD enabled for IS-IS but with very relaxed timers compared to IPv4. Haven't seen any issues for a couple of years.

Antti

----- On 18 Aug, 2020, at 15:11, Mark Tinka mark.tinka@seacom.com wrote:

> I'd recommend disabling IPv6 BFD for now as well. It's not handled in
> hardware, and so if it drops, you would lose your IGP. Worse if you run
> multiple address families in your IGP, e.g., IS-IS, or IPv4 over OSPFv3 .
>
> Hardware support for BFDv6 is coming in 1H'21.
>
> Mark.
>
> On 18/Aug/20 13:01, Mihai wrote:
>> Hi Ivan,
>>
>> Indeed, that was the issue, thanks!
>>
>> Regards
>>
>> On 18/08/2020 09:54, Ivan Malyarchuk wrote:
>>> If you have many uncontrolled directly connected L2 domains like
>>> IX-es or customers, check policer __default_arp_policer__.
>>> If there are drops, you need to apply interface-specific arp policers
>>> to interfaces with protocols and/or to source of arp bursts.
>>>
>>>
>>>
>>> 18.08.2020 02:35, Mihai ?????:
>>>> Hi,
>>>>
>>>> I have a MX204 with a couple of OSPF/OSPFv3 adj to other MXs over
>>>> two et- interfaces, BFD enabled for both protocols.
>>>> At random intervals the IPv4 BFD sessions are flapping with or
>>>> without the 'no-delegate-processing' option, however the IPv6
>>>> sessions are stable (they are using the same timers).
>>>>
>>>> Did anyone experienced something like this? The router is running
>>>> 18.2R3-S3.
>>>>
>>>>
>>>> Thanks!
>>>> _______________________________________________
>>>> juniper-nsp mailing list juniper-nsp@puck.nether.net
>>>> https://puck.nether.net/mailman/listinfo/juniper-nsp
>>>
>>
>> _______________________________________________
>> juniper-nsp mailing list juniper-nsp@puck.nether.net
>> https://puck.nether.net/mailman/listinfo/juniper-nsp
>
> _______________________________________________
> juniper-nsp mailing list juniper-nsp@puck.nether.net
> https://puck.nether.net/mailman/listinfo/juniper-nsp
>
>
> --
_______________________________________________
juniper-nsp mailing list juniper-nsp@puck.nether.net
https://puck.nether.net/mailman/listinfo/juniper-nsp
Re: IPv4 BFD flaps on MX204 [ In reply to ]
On 19/Aug/20 10:37, Antti Ristimäki wrote:

> Hi,
>
> In fact inline IPv6 BFD is supported for other than link-local addresses starting from 18.1 IIRC. This doesn't help for IS-IS or OSPFv3, though, as those use link-local addresses for adjacencies.
>
> We do have IPv6 BFD enabled for IS-IS but with very relaxed timers compared to IPv4. Haven't seen any issues for a couple of years.

If your timers are not aggressive, then you should be fine, as that will
account for any CPU spikes that de-prioritize BFDv6 packets.

We are running 150ms @ 3X and 250ms @ 5X. So we saw issues on BFDv6.

Mark.
_______________________________________________
juniper-nsp mailing list juniper-nsp@puck.nether.net
https://puck.nether.net/mailman/listinfo/juniper-nsp