Trouble understanding graph

[GrizzlyAK]

I have the following route in my network. The endpoint works as expected, but the route seems strange since it goes through a powered device (ZEN15) that is sitting a drawer behind me as I type this, IOW, this route is impossible. This image was taken just after rebuilding the route for the endpoint. Am I missing something? I know there has been much discussion for years on a graph, and I understand how difficult it is, given data availability and appropriateness of options vs size of the network. No criticism, just trying to understand what I'm seeing. Is the graph still a viable utility?

[robertsLando]

The best to answer this is @AlCalzone. My opinion is the graph is good the problem is zwave mesh protocol that sometimes does super weird things, it's not the graph that isn't reliable, we just represent what the protocol tells us about the network

[GrizzlyAK]

I do understand this, and know you can only do the best with what you are given. It's not a
criticism because I don't know how much zwave has or has not matured since I've backed away, so couldn't be sure if this was a graph fault or, like you say, an issue with zwave mesh reporting. Either way, I felt it prudent to bring it up since when I see something as glaring as this, I tend to question the efficacy of everything else around it. A network graph is an important feature (for me, at least). Cheers.

[AlCalzone]

I don't understand - why would the route be impossible if it goes through a powered device?
Since a while ago, we use the actual routes taken to draw the network map.

[GrizzlyAK]

😁 You must have missed the part that said the device is "sitting in a drawer behind me" UNPLUGGED. IOW, it is not plugged in nor is it participating in the network, and, no, it is not a battery powered device. It is the ONLY path indicated for the endpoint and the only way shown by the graph of communicating with the controller is THROUGH a non-existent device.

[AlCalzone]

Oof. Yes, I did miss that.

When did you last send a command to the device? Route information is only updated when outgoing communication happens. Z-Wave has no persistent connections and we don't the information about which routes were taken by the device to report back.

If you enable live mode on the network map and then ping the device, how does it look then?

[GrizzlyAK]

I thought you might have. ;-) The image posted was taken just after using the zwave js ui to rebuild the route for that device. However, I just checked it again, and it has a new route that does not include the dead device. Maybe I didn't give ZWave JS sufficient time to update its DB following the rebuild, or it has since figured it out. Either way, it now seems OK and there are no dead devices being used in any route. I have my controller on a powered USB hub, but around the same time as I noticed this issue, I moved it further away from my Pi and the graph now seems to show better connections and higher speeds with fewer hops. Maybe that, or just you guys working your MAGIC! I'll close this discussion. Thanks guys.

[AlCalzone]

Although it sounds weird, rebuilding routes does not automatically update the network map. This only happens when there is outgoing communication, which can take a while.

[GrizzlyAK]

Good to know! Thanks Again.

[GrizzlyAK]

Issue was resolved automagically!