[CALL FOR TESTING] PPPoE restructuring and IPv6 improvements

Started by franco, August 09, 2024, 09:11:31 AM

Previous topic - Next topic
Just updated.

I see no difference in PPP behaviour since unpatched 24.7.3
Still have to do the vlan-jiggle to get PPP to come up, still have to manually start dhcp6, but overall a smooth update.

@HCB thanks, looking at your mail at this very moment. Sorry, it's been too busy.

I use a dynamic IPv4 connection and use 6TO4 for dynamic IPv6. Is it a good idea to test the current patchset, or are there any known issues for this kind of setup?

PPP-type IPv4? 6RD and 6to4 quirks have been ironed out due to previous feedback. I don't expect them to break.

At the moment the only patch to test is the second one, because the first one is in 24.7.4 now:

# opnsense-patch https://github.com/opnsense/core/commit/6592a9a69

At the moment there seems to be a problem with PPP gateway setting in the mpd.conf which needs a hotfix. But it didn't come up in the testing here, possibly because the thread is mostly about PPPoE.


Cheers,
Franco

I have some feedback regarding PPPoE + native IPv6 by DHCP. I noticed that when (I think) ISP forces PPPoE session refresh then OPNSense fails to renew IPv6 DHCP lease. I have static /56  and from my understanding of logs ppp log seems to be ok, but "opnsense", dhcp6c and rtsold, in consequence are not happy. Way around it is to refresh WAN interface configuration. This is not related to recent release. It happened even on 24.7 but it happens very early (I noticed too few occurrences to figure out a pattern)

Aside from that PPPoE rework works great on every release for me.

This is a general issue when the DHCPv6 server forgets the lease or rejects it. Does it not auto-recover after 10-30 minutes? I'm not suggesting that's ok, just trying to get a baseline since I've worked on fixes for dhcp6c regarding this situation.


Cheers,
Franco

Quote from: franco on September 13, 2024, 04:29:53 PM
This is a general issue when the DHCPv6 server forgets the lease or rejects it. Does it not auto-recover after 10-30 minutes? I'm not suggesting that's ok, just trying to get a baseline since I've worked on fixes for dhcp6c regarding this situation.


Cheers,
Franco

Haha, internet is so essential that I was not patient enough to try to see if this fixes itself in up to 30 minutes :D At least when it happened during the day. Come to think of it I don't know if it ever happened during the night and I haven't noticed. I will try to monitor that more closely.

EDIT: Actually I remembered one day waking up to internet being MIA and once I checked the logs and confirmed just now the session refresh >2h earlier and it didn't fix itself until WAN config re-save.

Thanks, we may be missing something in that case. Maybe the interface index shifts for PPPoE or the device is otherwise unavailable because it was recreated but still referencing the old one. For now I made a ticket:

https://github.com/opnsense/dhcp6c/issues/39

Since we had to hotfix 24.7.4 in this area the second patch backport doesn't apply. Here is an updated version:

# opnsense-patch https://github.com/opnsense/core/commit/d24300f9ab


Cheers,
Franco

Since the second patch is in 24.7.5 now I'll unsticky this thread now. Thanks all for the help!


Cheers,
Franco