dhcp6c_script Runs Every Second

[990e468]

Hello

I have a VM running OPNsense which has two NICs passed through - one an intel 10Gbps SFP+ and the other an Intel 1Gbps NIC which is used for my WAN connection. My WAN connection is through PPPoE.

Ever since I upgraded to 24.1 I had issues where the router would stay up for around an hour before freezing and needing a reboot. I was able to SSH in and I noticed that top and ps would be full of

Code: [Select]

/usr/local/bin/php /usr/local/etc/rc.newwanipv6 pppoe0 forceAfter further inspection it looks like these were being run every second or so and never exiting, eventually exhausting system memory. I was able to work around this by having cron call the following script every minute

Code: [Select]

ps aux | grep -i php | grep -i rc.newwanipv6 | awk '{print $2}' | xargs killThis isn't ideal but has solved the stability issue.
Looking through the logs it looks like every dhcp6c_script runs every second (see attached log).

I could ignore it but I've also observed that ping times that are usually ~1ms jump up to 500ms roughly every 10 seconds, and WireGuard connections on client devices stop working until I disable them and restart them.

Is this something anyone else has experienced? Is there any way to diagnose this further?

Thanks

[cookiemonster]

Do you need ppoe on the VM? The WAN interface is likely to be set to ppoe on the configuration type, either or both v4 and v6, so the system keeps trying to get authenticated.

[990e468]

Thanks for your reply

The WAN interface is a PPPoE interface which runs on top of a VLAN. That VLAN's parent is the 1Gbps NIC I mentioned in the first post. That NIC is passed directly to the VM (i.e. it doesn't appear on the host system, only in the OPNsense router) using PCIe passthrough.
Using PPPoE over a tagged VLAN is a requirement for my ISP - PPPoE over VLAN 10 is pretty commonplace where I live (although it is rather annoying).

I've checked the "Point-to-Point" log file and the latest entries are from the 17th - the last time I rebooted my router. So I don't think it's an authentication issue.

The WAN interface is configured at PPPoE for IPv4 and DHCPv6 for IPv6. The "Use IPv4 connectivity" option in the DHCPv6 settings is checked, this is another requirement of my ISP.

[cookiemonster]

got it, it's just that the symptoms seem to point to the interface having to keep trying, for some reason.

[sillyabraham8@gmail.com]

Nice Post Thanks for sharing this post with us.

[cookiemonster]

According to https://github.com/search?q=repo%3Aopnsense%2Fcore+dhcp6c_script&type=code
that match the log at: 

Code: [Select]

2024-02-20T19:30:13 Notice dhcp6c dhcp6c_script: REQUEST on pppoe0 renewall (one of many), it makes me still think that for some reason the system gets a trigger to request a new renewal of the connection. Why that is, is the interesting question. I see no other clues there.
Maybe the ISP is having trouble, or something else. I don't use ipv6, sorry.

[990e468]

I did a packet capture and while it didn't show much in the way of PPPoE connections coming and going there were DHCPv6 requests being made every second. I haven't been able to tell why though - and it looks like all of them are getting status code "NoAddrAvail"

I did have an issue recently where I was getting IPv6 prefixes but no IPv6 address for the router. I enabled the "Request only an IPv6 prefix" and saved and applied and the router immediately kernel panicked. Since things are mostly working I might just leave this one be for now. I'll try updating to 24.1.2 later this week

Thanks again for your assistance

[depc80]

Do you need ppoe on the VM? The WAN interface is likely to be set to ppoe on the configuration type, either or both v4 and v6, so the system keeps trying to get authenticated.

It happened to me today, I tried create a cron work but DHCp6c renewal not stopping and quickly fill up memory usage.
Not sure if it has something to do with Zenarmor, was adding more in Policy.

[bronzeblood]

I am also getting the same issue, dhcpd6 seems to crash and run every second or so , which eventually fills up memory and crashes the router.

This seems to have started after i unchecked "Request only an IPv6 prefix" in wan dhcpv6 config, this was accidental but ever since i get this /usr/local/etc/rc.newwanipv6 script running every second.

It seems if the wan interface gets an ipv6 address the script runs over and over.