Jump to content

Archived

This topic is now archived and is closed to further replies.

tc60045

Add Upstream Monitoring Targets

Recommended Posts

tc60045

Really good monitoring for a residential network requires, in my opinion, upstream monitoring:

(a) of the Modem to which the network is connected

(b) to the nearest physical network "node" in the ISP's network, to test for connectivity.

 

I respectfully ask Domotz to consider adding these features.

 

For the first, in the US, Comcast / Xfinity customers (largest US broadband company) buy or lease modems that have a particular quirk:  they are accessible via 192.168.100.1 -- which is irrespective of the customer IP range (typically 10.x.x.x or 192.168.1.x or 192.168.0.x. and yes, many people use others... I get it).

 

This modem is addressable via the local network, as an http request to that address will be passed through the router to the upstream modem, which then responds.  

 

But Domotz will never find the modem on the local network -- unless it is told to look.   Add a telnet (port 80) or ping to that address as part of the network discovery process and you'll see the device.  More explicit interrogation is possible.  

 

The point of doing this is both to see whether the modem is responding properly and is online AND to keep the modem UP (in concert with my second request below).    The vexing "turn off internet on idle" settings on routers and modems can cause real headaches when diagnosing problems.   By pinging the modem, you ensure the router stays awake;  by pinging the first upstream node (below), you keep the modem awake.

 

For the second part of my request, when connectivity problems arise, the culprit may not be the user's network but rather feedback on the same cable "run" / "drop" in your neighborhood.  I spent six months helping Comcast ferret out a problem in our residential neighborhood, installing smokeping on raspberry PIs in neighbor's houses (and my own).  We found that a new neighbor had some very old TVs that were directly connected to the coax.  Even though he could get no unscrambled signals, he kept these TVs wired and used different inputs.  The TVs lacked rectifiers and flooded "back EMF" onto the coax and into the whole neighborhood.   Smokeping provided the diagnostics that led to technicians descending on this neighbor's home to fix connectivity for us all. 

 

So, in the monitoring section, it would be great to have Domotz do a few trace routes to confirm the first upstream node that *most likely* belongs to the ISP.  If a different node needs to be specified, let a user input this.  Or if that trace routing is hard, just let a user hand-define an upstream site.  Then ping test to that upstream node as often as you can, given all the other important stuff Domotz is doing.  When you see ping times go through the roof and stay high, you know you're having upstream problems.  That evidence will immediately convince your ISP that you know what you are talking about -- trust me, I was called "Mr. Smokeping" by the techs who started *CALLING ME* when they saw that I knew the problem was in our neighborhood.

 

 Most respectfully submitted

 

TC

Share this post


Link to post
Share on other sites

×
×
  • Create New...