[Solved] 18.1.5 issues

[elektroinside]

I don't know what's happening after the upgrade on my box.

So, here it goes:

1. Whenever I restart the box, I have no internet connectivity on the LAN clients; pinging from the OPNsense GUI works fine, pinging from the LAN clients (using IP or FQDN) fails
2. To make things work again on the LAN side, I have to either:
- disconnect/connect my PPPoE link (on the WAN)
- or edit the default gateway without any modification, save and apply
3. Right after the reboot, a lot of things are still loading of course, but the GUI is available at one point. When some of the services loaded (as pictured in the attached Screenshot_36.png), internet works on the LAN side. When everything is fully loaded (as pictured in Screenshot_37.png) internet on the LAN side no longer works
4. Sometimes I can't even ssh to the box from the LAN if I don't reconnect the WAN to fix the internet connectivity (something is not binding to some interfaces, I guess)

Errors in the log:
Line 63: Mar 22 19:40:49 gateway kernel: module_register_init: MOD_LOAD (vesa, 0xffffffff810ab110, 0) error 19
Line 97: Mar 22 19:40:49 gateway kernel: pcib0: _OSC returned error 0x10
and a bunch of "Line 268: Mar 22 19:40:54 gateway sshd[48206]: error: Bind to port 22 on ... failed: Can't assign requested address."

I already tried a clean install, which in my case is a pain in the *ss:
1. Install 17.7.5 first, because I get the segmentation fault error with 18.1
2. Upgrade to the latest version
3. Install plugins
4. Restore backup

[mimugmail]

Do you run some VPN in the Box? Disable gateway monitoring

[elektroinside]

I do have OpenVPN.
I didn't have gateway monitoring at the beginning. I enabled it later, to see what changes...
But having that disabled didn't work.
I'll try again...

Update: confirming that disabling gateway monitoring doesn't help... same thing

[Davesworld]

If you relied on the NAT patch, did you reapply it after the update? This fix will not be included until 18.1.6.

# opnsense-patch 7a823c56a

[elektroinside]

No, it's not my case. I think something changed in the network stack (just a hunch).
But thanks anyway.

[elektroinside]

Actually, a lot changed on the stack, judging by the changelog... will take the opportunity to underline (once more) the importance of a stable/better tested (by the community as well) release channel.

[franco]

Hi there,

18.1.5 has been unusually quiet and apart from PTI/IBRS lacks behavioural changes almost entirely.

Just try with 18.1.4 package to see if that caused it:

# opnsense-revert -r 18.1.4 opnsense

The only change that looks related is "interfaces: ignore dynamic linkup events for unassigned interfaces".

If it's the same issue, I don't particularly believe it's the kernel but why not switch back this one as well...

# opnsense-update -kr 18.1
# /usr/local/etc/rc.reboot

It could be a local issue surfacing more clearly and / or entirely unrelated to the software. Reinstall does not yield different results most of the time.


Cheers,
Franco

[elektroinside]

Yep, 18.1.4 works fine after a reboot. Rebooted 3 times, no issues.
The downgrade was performed in a matter of seconds. At first, i believed it failed because of the speed, but the GUI is showing that I'm on 18.1.4. So i think it worked. Anyway, rebooting no longer breaks internet connectivity on the LAN side, so there's something probably incompatible with my @$&43#_-;!!?: PPPoE link and OPNsense 18.1.5.

I really like the new kernel. Despite the slightly higher CPU usage, it is very snappy and everything is just very fast.

[franco]

Sorry for the late repy.

https://github.com/opnsense/core/commit/d94ed6d8

Flip this... either from 18.1.4 or 18.1.5. 18.1.4 would get worse, 18.1.5 would get better when done.

# opnsense-patch d94ed6d8

(It's a theory.)


Cheers,
Franco

[elektroinside]

Updated to 18.1.5 and applied the patch.
And it works!

Thank you Franco, well done!

Btw, it's "opnsense-patch", right? :)

[fabian]

yes

[franco]

Alright, this is a fix so something else is going on... what is your WAN and LAN config with regard to IPv4 and IPv6 setup types?


Cheers,
Franco

[elektroinside]

WAN: IPv4 is PPPoE, IPv6 is DHCPv6
LAN1: IPv4 is "Static IP", IPv6 is "Track interface"
LAN2: IPv4 is "Static IP", IPv6 is disabled (this is for the guest wifi network, without a captive portal)

On both LANs, DHCPv4 is running, no DHCPv6.
Simple, by the book setup :)

[franco]

When you say:

"1. Whenever I restart the box, I have no internet connectivity on the LAN clients; pinging from the OPNsense GUI works fine, pinging from the LAN clients (using IP or FQDN) fails"

Do you mean LAN1, LAN2 or both?


Cheers,
Franco

[franco]

And more interestingly: Is IPv4 or IPv6 not working, or both?