I have been experiencing numerous lockups lately. I have seen it happen on all of the recent OS versions, including 2.9.14 that I am currently running.
The symptoms:
Everything works perfectly until approximatly 200 users are logged on (meaning they show up in the MT Interface list). Then suddenly…
Users attempting to log on can not.
Users already connected suddenly can not pass any data
SNMP traffic graphs show a sudden drop in throughput
The only recourse I have found is to reboot the MT
The configuration:
All users connecting over copper DSL, running either RasPPPoE on their PCs, of their DSL modems are configured as routers and they perform the PPPoE logon.
The DSLAM is connected via ethernet to an ethernet interface on the MT router.
The MT PPPoE server is using radius for authentication against a remote radius server.
Whenever this happens, customers start calling in and screaming, so my support staff race over to the MT and reboot it. I finaly set up remote logging so I could see the logs at the time this happens, even if they reboot the MT before I can look.
And sure enough, there they were. Log after log of “200 limit exceeded”. Boy, I don’t remember that limit on PPPoE connections when I bought the MT. Is this something that came with one of the updates? I see a mention on the web site that as of version 2.10, PPPoE will be limited to a single connection for level 3 licenses! Ouch, that would have hurt me.
So it looks like my only recourse is to buy a level 6 license which is unlimited PPPoE connections. I have an email into MT now for the quickest way to upgrade my license, this is really frustrating my customers until I can get this resolved.
You should have at least 2 MT’s in place. The PPPoE protocol is Layer 2 ethernet and not layer 3 IP based. Thus it is inherently simple to add a second box for load sharing redundancy…