PPPoE connection in Opnsense stopped working

[Ilitirit]

(Originally posted here, but this might be a better forum)

For some reason I can no longer connect to my ISP (Cell C/Openserve) via my Opnsense router. It was working fine for months. I can connect without problem directly via Windows or using a different router.

These are the logs from a login attempt:

LCP: SendConfigAck #108
  MRU 1492
  AUTHPROTO PAP
  MAGICNUM 0x2c6160ec
LCP: rec'd Configure Ack #93 (Ack-Sent)
  MRU 1492
  MAGICNUM 0xa9f48fd4
LCP: state change Ack-Sent --> Opened
LCP: auth: peer wants PAP, I want nothing
PAP: using authname "[email protected]"
PAP: sending REQUEST #1 len: 44
LCP: LayerUp
LCP: rec'd Configure Ack #93 (Opened)
  MRU 1492
  MAGICNUM 0xa9f48fd4
LCP: state change Opened --> Req-Sent
LCP: LayerDown
LCP: SendConfigReq #94
  MRU 1492
  MAGICNUM 0xa9f48fd4
LCP: rec'd Terminate Request #109 (Req-Sent)
LCP: SendTerminateAck #95
LCP: rec'd Terminate Request #109 (Req-Sent)
LCP: SendTerminateAck #96
PPPoE: connection closed
Link: DOWN event
LCP: Down event
LCP: state change Req-Sent --> Starting
Link: reconnection attempt 26 in 3 seconds

Anyone know what would cause this behaviour?

 

[Ilitirit]

I've come across posts on forums where people with the same problem are saying that it's due to a configuration change on the host side that Opnsense can't/doesn't handle yet. Could be the addition of an echo requirement or similar. Not really sure about the details, and I can't really verify it any way. In any case, I'm going to see if I can get a build of pFSense that's compatible with my router hardware and check if it exhibits the same behaviour.

 

[Ilitirit]

After some more testing and research I'm relatively sure these were the causes:
1. The addition of or change to Openserve's echo requirements
2. A limitation(?) in Opnsense (FWIW, I experienced the same issue when testing in pfSense)

One of the reasons it was hard to detect is that by default Opnsense doesn't log Echo Requests (or it gets DC'd before it can log them?). If Openserve doesn't receive a response in X amount of seconds, you get disconnected. It seems that at some point during PPP connection/negotiation Opnsense is blocking and queueing messages in order to run other processes. This prevents it from responding in time which causes Openserve to send a termination request (see #109 in the log snippet above).

Opnsense detects the disconnection and tries to reconnect, leading to an infinite loop which in turn presumably led to the connection "ban" mentioned in the link in the original post. My router was probably hammering their server with connection attempts - possibly by multiple simultaneous processes.

[To confirm it was not a hardware issue, I connected from a shell using ppp and very basic configuration. The connection succeeded and was stable. The Echo Requests and Responses were clearly visible in the log.]