Bug #14434
open
PPPoE WAN interface with VIPs causes continuous interface restarting
0%
Description
I have a /28 routable legacy IP block from the ISP, and they assign the first usable address of the /28 block as a /32 to the PPPOE interface, so i have:
Routable block: x.x.x.64/28
PPPOE address: x.x.x.65/32
LAN address CARP VIP: x.x.x.65/28
This configuration worked fine in 22.05, but is broken in 23.01 and remains broken in 23.05.
The PPPOE connection establishes and calls /etc/rc.newwanip, which then calls find_interface_ip() and get_interface_ip() to determine the address assigned to pppoe0. These functions return NULL, which causes rc.newwanip to restart the pppoe0 interface. This then causes an endless loop. The logs show the correct interface name, but no IP:
rc.newwanip: on (IP address: ) (interface: WAND[opt5]) (real interface: pppoe0).
Looking through the find_interface_ip() function, i can see it looks for $interface_ip_arr_cache - this array exists, but is empty causing the function to fail and return NULL.
I can see that if $interface_ip_arr_cache does not exist, it should open /var/db/${interface}_ip
if (!isset($interface_ip_arr_cache[$interface]) or $flush) {
if (file_exists("/var/db/${interface}_ip")) {
The file /var/db/pppoe0_ip is present and contains the correct address.
I'm hoping someone more familiar with the codebase and changes between 22.05/23.01 could give some insight into this otherwise i'll be trying to track it down further.
Updated by Steve Wheeler over 1 year ago
- Project changed from pfSense Plus to pfSense
- Subject changed from PPPOE WAN interface overlapping with CARP VIP causes continuous interface restarting to PPPoE WAN interface with VIPs causes continuous interface restarting
- Category changed from PPP Interfaces to PPP Interfaces
- Priority changed from Normal to High
- Target version set to 2.8.0
- Affected Plus Version deleted (
23.01) - Plus Target Version set to 23.09
- Affected Version set to 2.8.0
This also affects 2.7 and when using IPAlias VIPs on the WAN.
See: https://forum.netgate.com/topic/181345/2-7-0-pppoe-not-working
Updated by Steve Wheeler over 1 year ago
- Affected Version changed from 2.8.0 to 2.7.0
Updated by Jim Pingle about 1 year ago
- Plus Target Version changed from 23.09 to 24.03
Moving the target ahead for now but there have been several other fixes for interface/VIP functions in 23.09 already so it's possible the underlying problem has already been addressed.
Updated by Adam French 10 months ago
Steve Wheeler wrote in #note-4:
Still present in 23.09.1
I can also confirm it is still present is the latest stable release since I was affected by this issue too.
Oddly removing the VIP's has not affect for me though. PPPoE comes up for very short period and goes back down again. Same log events saying "caught fatal signal TERM"
When monitoring this I noticed the line does come up and is assigned an IP using DHCP from our ISP, but quickly get's shutdown when this log event occurs.
This was previously working in 22.05.1, stopped in 23.01 and continued to not work in 23.05 and also in 23.09.1
Whatever change was made between 22.05.1 and 23.01 has broken the PPPoE functionality.
Updated by Jim Pingle 9 months ago
- Plus Target Version changed from 24.03 to 24.07
Updated by Pete Holzmann 6 months ago
I thought I would add another confirmation: I have the same symptoms on upgrade from 2.6.0 to 2.7.0
I too am using PPPoE with VIPs to provide HA CARP on our WAN
Happy to do what I can to help.
I am now using 2.6 on main and have 2.7 on backup CARP, in a largely homelab environment...
so as long as my sweetie doesn't have something urgent I can run tests.
Updated by Steve Wheeler 6 months ago
This appears to be fixed in 24.03. At least in my test setup. Can anyone who was seeing this in 23.09.1 confirm that in a real world setup?
Updated by Jim Pingle 6 months ago
- Plus Target Version changed from 24.07 to 24.08
Updated by Jon Gerdes 2 months ago
Steve Wheeler wrote in #note-8:
This appears to be fixed in 24.03. At least in my test setup. Can anyone who was seeing this in 23.09.1 confirm that in a real world setup?
I have a CE 2.7.2 two node CARP cluster in production. Two of its WANs are PPPoE.
I created a separate interface on each node for the physical NIC that is dialed through, with an RFC1918 address for each interface. A single CARP address in the same RFC1918 subnet uses those real IPs. Multiple external IP aliases use the CARP address as their parent, so the PPPoE link itself gets the first usable address in the /28 or /29 and the IP aliases do the rest.
They have been as stable as the ISP 8)
Updated by Jim Pingle about 2 months ago
- Plus Target Version changed from 24.08 to 24.11
Updated by Marcos M about 1 month ago
- Target version deleted (
2.8.0) - Plus Target Version deleted (
24.11)