IPv6 wireless issues

[jared@…]

I'm seeing periods of time where IPv6 goes unavailable on wireless. This seems similar to what was seen at Chicago/IETF98 where the prefix lifetimes were less than prefix intervals.

[jared@…]

see https://tickets.meeting.ietf.org/ticket/1095

for the prior issue.

- Jared

-- 
Jared Mauch  | pgp key available via finger from jared@puck.nether.net
clue++;      | http://puck.nether.net/~jared/  My statements are only mine.

[jim@…]

Spoke with Jared. He believe's he was on ietf, but isn't certain. The claim is that the prefix lifetime in the RA is shorter than the unsolicited RA interval. Please check both the Junipers and the Edgerouters to ensure it's set reasonably.

[Clemens Schrimpe]

... that would be embarrassing - will 👀

[jared@…]

I'm fairly sure I was on ietf-hotel while in the meeting space based on
my saved ipv6 address in my buffer, which was: 2001:67c:1232:144:8e:3d97:857d:8d32

[jim@…]

Looks good on the Junipers. From the group templates at the top of both:

RA_Config {

protocols {

router-advertisement {

interface <*> {

max-advertisement-interval 5;
min-advertisement-interval 3;
other-stateful-configuration;
virtual-router-only;
default-lifetime 600;
dns-server-address 2001:67c:370:229::6;
dns-server-address 2001:67c:370:229::7;

}

}

}

}

[Clemens Schrimpe]

On the indicated network (2001:67c:1232:144::/64 → "ietf-hotel" wireless network) the prefix lifetime is (and was) one month → 2592000 seconds.

If you keep losing connectivity we'll have to investigate - there might be different reasons.

[llynch@…]

push for email update (changes in trac)

• Lucy

[jim@…]

Jared, could you please come by the NOC and work with Clemens if this reoccurs? We've been unable to reproduce.

• Jim

[Clemens Schrimpe]

As per agreement w/ Jared we are closing this ticket now. He'll re-open it if he ever encounters The Problem™ again.

[Rick Alfvin]

Milestone renamed