Hell of a necro post, apologies for that, but this is exactly the place to ask my question.
I have a setup
RB2011-RM serving as a PPPOE server (Radius behind that)
40 CPE's connected to the RB2011-RM (Hap Lite's)
Each CPE has 2 PPPoE clients set up, 1 is for internet, 2nd is for management
What I have found is that if the RB2011-RM somehow loses power, the CPE's maintain their PPPoE session.
The RB2011-RM restarts and waits for new requests (All previous sessions being closed on the RB2011) but doesn't get any because the CPE still thinks it has an active working session (2 sessions per CPE).
In effect the CPE never closes its session and thus never tries to reconnect.
CPE PPPoE client settings
Keepalive timeout is blank (Default)
on the Profile
Session Timeout = blank (Default)
Idle Timeout = blank (Default)
RB2011 PPPoE Server settings
Keepalive timeout = 60
One Session per host = NO
Max Sessions = blank
on the Profile
Session Timeout = blank (Default)
Idle Timeout = 30 Minutes
Only One - NO
Would greatly appreciate your feedback.
Sorry for reviving your necro response to a necro post
I just happen to be troubleshooting a small issue with PPPoE disconnects too, and stumbled upon this post. Did you ever manage to solve this?
From what you tell in your post, what I gather is that it should be the
clients that implement a keepalive in this situation, right? You had only implemented a keepalive on the server side (the RB2011), but that keepalive is never triggered because it's the RB2011 that actually loses power. On the CPE side, the tunnel is never closed either, because there's no keepalive. They don't care if the RB2011 goes down, they just keep the tunnel up.
Wouldn't this be solved by just setting a keepalive in the CPEs too? That way if the RB2011 doesn't respond for more than X seconds, the CPEs will consider the PPPoE tunnel dead and try to reconnect.
Hopefully this is how you solved it... as I'm going to do the same thing on my end
