IPv6, RA and static IPv6 assignment
Are there any plans to allow the probe's listening for RA to be disabled so that I can configure a static IPv6 address, router and DNS server without RA overwriting my values?
mgm> Are there any plans to allow the probe's listening for RA to be mgm> disabled so that I can configure a static IPv6 address, router and DNS mgm> server without RA overwriting my values? I am a teensy litttle bit worried about the missing green checkmarks in the probe management console, too. But it is mostly that, I guess. A related issue, not mentioned on this list so far: I'm v6-multi-homed at home with 2a00:1030:100::/48 2a00:1030:1004:1000::/56 and RA-announce two prefixes/gateways on the probe's net. It's a matter of luck whether the probe auto-configures its primary address to the prefix I'd prefer (the one also used in the static prescription). (That's not specific to the ATLAS probe -- a lot of my gear is non-deterministic in this respect.) To get specific: My probe 2781 has a static configuration for: 2a00:1030:100::149 It currently uses as its primary address: p2781.probes.atlas.ripe.net has IPv6 address 2a00:1030:1004:1000:220:4aff:fee0:2212 In fact, it responds to its one static and two autoconfigured addresses. You can ping6 any of these: 2a00:1030:100::149 2a00:1030:100::220:4aff:fee0:2212 2a00:1030:1004:1000:220:4aff:fee0:2212 I'd happily trade in the two extra SLAAC adresses for a single extra static address: 2a00:1030:1004:1000::149 (You'll find the DNS for atlas.ml.gaertner.de already prepared for that. That also means that only every other ping6 on the name succeeds. :-) I'm not complaining, just pointing out actual behaviour here at my little home lab. The ATLAS project is big fun and that the RIPE ATLAS team is doing a great job! Martin
On 2013/09/13 22:27 , neitzel@marshlabs.gaertner.de wrote:
A related issue, not mentioned on this list so far:
I'm v6-multi-homed at home with
2a00:1030:100::/48 2a00:1030:1004:1000::/56
and RA-announce two prefixes/gateways on the probe's net. It's a matter of luck whether the probe auto-configures its primary address to the prefix I'd prefer (the one also used in the static prescription). (That's not specific to the ATLAS probe -- a lot of my gear is non-deterministic in this respect.)
At my home I have 4 IPv6 prefixes (2 from xs4all, HE, and 6to4). The best way to retain some sort of sanity is to make sure that all but one are deprecated. Alternatively, make sure that all just work and that it doesn't matter which one is used. The main problem there is reverse DNS. Philip
A related issue, not mentioned on this list so far:
I'm v6-multi-homed at home with
2a00:1030:100::/48 2a00:1030:1004:1000::/56
and RA-announce two prefixes/gateways on the probe's net. It's a matter of luck whether the probe auto-configures its primary address to the prefix I'd prefer (the one also used in the static prescription). (That's not specific to the ATLAS probe -- a lot of my gear is non-deterministic in this respect.)
At my home I have 4 IPv6 prefixes (2 from xs4all, HE, and 6to4). The best way to retain some sort of sanity is to make sure that all but one are deprecated.
Alternatively, make sure that all just work and that it doesn't matter which one is used. The main problem there is reverse DNS.
my home is also IPv6 multi-homed. I expect the RIPE probe to probe using both source addresses. my routing infrastructure does SADR, so the probe by choosing source address also chooses exit. cheers, Ole
On 2013/09/19 12:21 , Ole Troan wrote:
At my home I have 4 IPv6 prefixes (2 from xs4all, HE, and 6to4). The best way to retain some sort of sanity is to make sure that all but one are deprecated.
Alternatively, make sure that all just work and that it doesn't matter which one is used. The main problem there is reverse DNS.
my home is also IPv6 multi-homed. I expect the RIPE probe to probe using both source addresses. my routing infrastructure does SADR, so the probe by choosing source address also chooses exit.
At the moment, probes do not explicitly select sources addresses. What gets used in a measurement is whatever the Linux kernel on the probe considers the best address for a particular destination. That may change in the future because source address selection may be needed for Atlas anchors, but at the moment the time frame is unknown. Philip
participants (4)
-
Mike. -
neitzel＠marshlabs.gaertner.de -
Ole Troan -
Philip Homburg