PPPoE flapping

[JaapHoetmer]

Dear all, At one customer I am experiencing DSL performance issues, and the ISP says the PPPoE session is flapping due to requests from the firewall (Fortigate 40C) to disconnect the PPPoE session. Can this be possible, and under what circumstances would the Fortigate send these requests? Fortigate 40C runs 5.0, build 0179 (GA Patch 2) The ISP is furthermore suggesting to establish the PPPoE session from their equipment, rather than from the firewall, as they say that may help to stabilise the connection. Do you have any experience with this? Any ideas or help will be appreciated!

[Dave_Hall]

Could be a duplex or speed mismatch issue. Perform a " diag hardware deviceinfo nic <interface>" on the wan port and check for errors. You can try setting the duplex/speed on the interface from the CLI. eg.

 # config system interface
 
 (interface) # edit wan1
 
 # set speed ?
 1000full    1000M full-duplex
 100full     100M full-duplex
 100half     100M half-duplex
 10full      10M full-duplex
 10half      10M half-duplex
 auto        auto adjust speed
 
 # set speed 100full
 
 # show
 config system interface
     edit " wan1" 
         set vdom " root" 
         set mode dhcp
         set allowaccess ping fgfm
         set type physical
         set defaultgw enable
         set speed 100full
     next
 end
 

Edit: never had to use this myself but the handbook does list other PPPoE related options that can be used on the interface, including a setting for the " idle-timeout" for PPPoE connections.

[ede_pfau]

You might have a look at the WAN interface Dead Gateway Detection parameters. If the defined ping server is unreliable the FGT will shut down the line by itself. If DGD is configured try again without having it enabled.

[JaapHoetmer]

Ok thanks gentlemen, I will have a look. I also noticed the following messages when enabling debug on ppp using diag debug application ppp 3: id=0 msg=" pppd is started" Couldn' t get channel number: Transport endpoint is not connected id=0 msg=" pppd is exiting" What do these messages indicate? They are intermixed with loads of these echo request-reply messages: PPP send: LCP Echo_Request id(113) len(8) [Magic_Number de1edd85] PPP recv: LCP Echo_Reply id(113) len(8) [Magic_Number 1c27cb60] PPP send: LCP Echo_Request id(60) len(8) [Magic_Number be6be82e] PPP recv: LCP Echo_Reply id(60) len(8) [Magic_Number 87212a78] which is probably a keepalive process from the ISP? Thanks for your help

[emnoc]

Yes I' ve seen this with my DSL provider ( ATT fastDSL ) and after they change something in their DSLAM. What you need to do is to modify the lcp echo interval under the interface via the CLI set lcp-echo-interval I can get you an extact number, but modify the setting time you get stability. Ken

[JaapHoetmer]

Thanks for the help, Ken. I have not been able to test this as the ISP has swapped the modem for a new one, and set the modem as the PPPoE initiator. The line is now stable but it disallows to determine the cause or causes of the problems. The ISP claims it is the firewall that decides to terminate the PPPoE session, the firewall says there' s no response to the echo requests.

PPP send: LCP Echo_Request id(37) len(8) [Magic_Number 5a6174eb]
 PPP recv: LCP Echo_Reply id(37) len(8) [Magic_Number 499a0eb3]
 PPP send: LCP Echo_Request id(15) len(8) [Magic_Number 68a55e53]
 PPP send: LCP Echo_Request id(38) len(8) [Magic_Number 5a6174eb]
 PPP send: LCP Echo_Request id(16) len(8) [Magic_Number 68a55e53]
 PPP send: LCP Echo_Request id(39) len(8) [Magic_Number 5a6174eb]
 No response to 3 echo-requests
 Serial link appears to be disconnected.
 PPP send: LCP Terminate_Request id(4) len(23) 
 PPP recv: LCP Terminate_Ack id(4) len(23) 
 Connection terminated.
 Connect time 1.5 minutes.
 Sent 147 bytes, received 1949 bytes.

Are there any known issues with PPPoE?