Hi Joel,
This behavior is not unspecified. It is specified in Sec. 4 of the draft
(as mentioned in my first email). Given the same packet and the same router
configuration, the behavior should be the same with all implementations.
Thanks,
Francois
On 7 Sep 2023 at 16:32:06, Joel Halpern wrote:
Why is it unspecified what the behavior is if the arg bits are
non-zero. Leaving things unspecified (and getting different behaviors
from different implementations) makes things more complex if we need to
build on this behavior.
Yours,
Joel
On 9/7/2023 10:30 AM, Francois Clad wrote:
Hi Joe
Hi Joel,
Sec. 10.1 specifies how the SR source node should send the ICMP packet so
that it is correctly processed by the receiver. I.e., if the source sends
an ICMP packet to a SID with a non-zero argument, then they may not get a
reply back, as you have observed in your tests.
Thanks,
Francois
Can you explain then what the text is section 10.1 is talking about? It
seems like it is mandating certain bits being 0 for it to be considered
a match? (I will also have to find out more about why the tests showed
certain routers dropping such packets.)
Yours,
Joel
On 9/7/2023 4:09 AM, Fr
Hi Joel,
The receiver in this case is an SR segment endpoint node processing an IPv6
packet that matches a FIB entry locally instantiated as a SID of this
document. Therefore, the expected behavior of the receiver is that
described in Sec. 4.
Thanks,
Francois
On 6 Sep 2023 at 16:38:30, Joel Hal
Askign as a participant:
In section 10.1 it says:
When the SRv6 SID in the destination address of the ICMPv6 echo request
is one of the SID flavors defined in this document, the SR source node
MUST set the arguments of the SID to 0
What is a receiver required to do if it gets an improper ICM
