192.168.100.1

For every one.
Makes perfectly good sense.
Your local gateway / router KNOWS that the .100 net is not "local".
So it passes the packets to its next gateway, which happens to reside on the other end of the modem.
Even if the local cable modem allows those packets through, the next step gateway at the ISP will NOT route them, they are illegal.

The modem KNOWS that anything starting with 192.168.100.1 is for it, grabs them, and allows the local connection.

The only way this whole scenario does not work is if you have a local .100 network, which means you 1st hop router won't pass the packets to the modem.

And since the cable company will not support local end-users with home networks other than trivial connection test, it is not their problem.

If you, on the other hand have setup a "non-standard" home network (whatever that means) with multilpe segments, etc, then you better KNOW what segments you assigned and what breaks what.

Note: This is for local log viewing and checking, so it is not like it really hurts something if you can't get to it, unless the cable guy shows up, and then he will tell you to plug in into a single DHCP enabled pc anyway.

It is working exacly as it is design to.
The cable modem is not ON your network, it is all by its lonesome like it should be.


You have pegged the downtime to bad weather.
Next time it happens, go to the cable status logs, grab screen shots (or wget pulls) od the messages concerning communication failures.

Actually, you might want to script a wget to pull them on a schedule.

They only keep the last 'x' entries. You would then write a script to dedupe the overlaps.

Then when you realize there is a LOT more downtime than your interactive usage notices, you can use the logs as proof to your cable company of how bad the service is, and force them to do a bit of rewiring.