Still having PPPoE Problems

Products ePMP
1

For about a year now , and though firmware versions and combination of firmware versions (currently 3.4.1 on AP's and 3.5 on SMs)  we have continued to have a problem were the SM's appear to stop attempting to connect via PPPoE. They just give up after a few tries and refuse to try again until rebooted or kicked off the AP and forced to reconnect.

We can reproduse the problem and log it now -

Using LVL 2 firewall rule on the AP I created a rule to do nothing but log PPPoE discover packets on the WLAN. 

Name: Log PPPoE Discovery on WLAN

Action: Accept

Log: ON

EtherType: 8863

I log into the switch at the site and turn off the ethernet port the AP is connected to so the radios stay connected to the AP but can not reach the PPPoE server. Wait about about 5 minutes (or less or more).  I turn the port back on. About 10 of the 15 clients that were on the AP have PPPoE discover packets show up in system log of the AP  instantly and they auth and work as expected. However, there will always be several radios (it's random, not always the same ones or the same number of radios though some seem more prone that others)  that no discovery packets show up.

This morning I waited over an hour and the 5 radios that were not authenticating never authenticated.  So then I went the AP and Kicked one off the AP. The second it reconnected -

Sep 1 01:38:02 APe65GS0 L2_Firewall: INPUT IN=ath0 MAC source = 00:04:56:ff:f3:b9 MAC dest = ff:ff:ff:ff:ff:ff proto = 0x8863
Sep 1 01:38:02 APe65GS0 L2_Firewall: FORWARD IN=ath0 OUT=eth0 MAC source = 00:04:56:ff:f3:b9 MAC dest = ff:ff:ff:ff:ff:ff proto = 0x8863
Sep 1 01:38:02 APe65GS0 L2_Firewall: FORWARD IN=ath0 OUT=eth0 MAC source = 00:04:56:ff:f3:b9 MAC dest = cc:2d:e0:1e:79:62 proto = 0x8863

The radio authenticated and worked as expected.  One after the other I kicked each of the remaining 4 off the AP and as soon as they reconnected their PPPoE discovery packet was logged and they authenticated.

Everything seems to point at the customer radios simply stop trying however it seems odd that we are the only ones experiencing this if that is the case.

2

Submit a ticket with your full logs and packet captures.

No firewall in the mix here?

3

It is known defect with PPPoE daemon on SM. It is already fixed in firmware version 3.5.5-RC8. Soon it will be available for the download on the Cambium support web page.

1 Like
4

WOOT ! 

5

Any idea when 3.5.5 is coming out to fix the pppoe issue?

I got two towers with Force200 clients and ePMP2000 Ap's havng the same issue.

6

"

 
Cambium Employee
 
‎11-26-2018 01:52 AM
Re: Still having PPPoE Problems

It is known defect with PPPoE daemon on SM. It is already fixed in firmware version 3.5.5-RC8. Soon it will be available for the download on the Cambium support web page."

Just FYI  this was either not completely fixed or broken again in v3.5.6  .   We had a 4 PoP's lose connection to our NOC for about 10 minutes this morning resulting in about 400 customers radios that could not reach the PPPoE server. This means they all lost their PPPoE session and would have been unable to reach the PPPoE server for about 10 minutes.   

Now before the aforementioned fix this would have resulted in 400 customers radios that just stopped trying to auth and thus we would have had 400 customer radios with hung PPPoE clients to deal with.  However this morning out of the 400'ish customer radios only 6 failed to authenticate when their path to the PPPoE concentrator was restored.  So, very much improved but not completely fixed. (there were 8 customer radios that had escaped updates and were still running v3.5 and the PPPoE client wasn't working on all 8 as a result but the 6 running 3.5.6 was a surprise).