Feature Request: Custom Clients, Topology Control, and Availability Monitoring

[Itay1787]

Hi,

I would like to ask if it’s possible to add support for creating custom clients, instead of relying only on what NUT automatically detects or assumes.
Also be great if custom clients could have an option to run continuous ping checks for status indication

In addition, in the topology view, I would like to be able to manually define or change:

What is considered the "data line" or "Active electrical line" between UPSs

And a clear indication of whether the UPS is currently reachable/available

At the moment, when a UPS becomes unavailable or disappears, its statistics simply freeze, which makes it hard to understand whether the UPS is offline or just not updating.

I believe these features would greatly improve flexibility, visibility, and monitoring accuracy.

Thanks in advance, and great work on the project!

[SuperioOne]

Hi,

Thank you for your feature request, but I need some extra explanation about some of the requests.

I would like to ask if it’s possible to add support for creating custom clients, instead of relying only on what NUT automatically detects or assumes.
Also be great if custom clients could have an option to run continuous ping checks for status indication

Creating a custom client is possible, but "client" in NUT refers to a server/PC that actively depends on an UPS for electrical power. These clients are meant to be used for deciding shutdown order by NUT. I'm not sure about the use case for creating a custom client solely to check UPS status since nut_webgui already does this within every poll_freq (2 seconds by default).

If the custom client is meant for running some custom action on a certain UPS event, yes its planned although upsmon already does a great job. I did some of the ground work and only remaining part is implementing a proper embedded language such as Lua or Wren.

In addition, in the topology view, I would like to be able to manually define or change:

What is considered the "data line" or "Active electrical line" between UPSs

Data lines are USB/SNMP/GPIO connections used for reading UPS information by NUT, and electrical lines are self-explanatory. The topology graph provides a quick overview of how NUT servers view the system, such as which computer is protected by which UPS, etc. Why do you need to manually change this?

(Just a note: There is currently a visual issue on the graph where both data line and electrical lines overlap on NUT server nodes.)

And a clear indication of whether the UPS is currently reachable/available
At the moment, when a UPS becomes unavailable or disappears, its statistics simply freeze, which makes it hard to understand whether the UPS is offline or just not updating.

This part might be a bug that I'm not aware of. How long are the statistics frozen?

By the nature of the poll-based system, there is a delay between UPS, NUT, and nut_webgui, but it shouldn't be longer than poll_interval + page poll. After that, if the UPS device no longer exists in NUT protocol responses, nut_webgui also removes it from the list.

If that's the case, I can look into the polling mechanism used for UPS disconnect/connect events (currently controlled by the poll_interval value) to make it sync faster with the NUT server.

[Itay1787]

Hi, thanks for the reply.

Creating a custom client is possible, but "client" in NUT refers to a server/PC that actively depends on an UPS for electrical power. These clients are meant to be used for deciding shutdown order by NUT. I'm not sure about the use case for creating a custom client solely to check UPS status since nut_webgui already does this within every poll_freq (2 seconds by default).

Ok, yes, my situation is maybe a bit different from the usual use of NUT.
I have 2 NUT servers that all talk to a main NUT server (in the end, 3 servers in total)
On the main server, I have scripts that run according to the status of the UPSs instead of upsmon (upsmon runs the main script, and from there it splits according to the status)
This gives me more flexibility and options if I need to wait for server A to shut down before server C, and shut down in a certain way, in addition to making sure that the shutdown was really done, etc, etc
And that's why I would be happy if I had the option to add those servers and services in the UI just for a quick index of their status.

Because currently, the server/pc that NUT pullers are not accurate for my use.

Data lines are USB/SNMP/GPIO connections used for reading UPS information by NUT, and electrical lines are self-explanatory. The topology graph provides a quick overview of how NUT servers view the system, such as which computer is protected by which UPS, etc. Why do you need to manually change this?

For the same reasons I mentioned above, but in addition to that, I don't know how it decides exactly what is Data lines are and what the electrical lines are? Because sometimes it's just the other way around, even though I connected to each server separately (thanks for this addition!)

This part might be a bug that I'm not aware of. How long are the statistics frozen?

Oh, ok, I was sure this was normal behavior.
I will mention that I just updated from 0.54 to the newest version today, maybe that will solve it at all
But as soon as the UPS disappears, its statistics simply remain frozen until it comes back, and then it refreshes with the latest info.
For example:
The UPS turns off at 37 percent, so in the UI it will remain at 37 percent, then the power will come back, and then suddenly it will jump to 0 percent when NUT reconnects to the UPS and new info is available. In no case does it disappear or do anything else.
On the contrary, I would be happy if, instead of it disappearing from the list, it would simply turn it gray or something like that to indicate that it is unavailable.

[SuperioOne]

I see, custom scripts have to have some sort of mechanism to report their existence and status. Normally, the NUT protocol provides this functionality via the LOGIN and PRIMARY commands. Either your own custom client has to have a similar mechanism, or I have to provide a custom client/client template, which is technically recreating upsmon v2.

I won't close this issue as not planned, but it's a little bit out of the scope for now.

I don't know how it decides exactly what is Data lines are and what the electrical lines are?

Data line: If the NUT server reports an UPS, it means it can communicate via some type of driver. These are listed as data lines.
Electrical: The NUT server reports primary and secondary client IPs attached to the specific UPS device. In general, these clients are upsmon, and they are mainly used for describing if the client depends on a UPS for power.

Of course, this information does not guarantee that there is a physical electrical cable between the UPS and the client since you can create arbitrary clients with raw TCP connections and upsmon.

If some of the lines look way off, it’s either that upsmon is attached to an UPS when in fact there is no physical connection at all, or there might be a dummy repeater driver that reports 127.0.0.1 as a client, which makes it hard to tell exactly which server it's attached to.

For example:
The UPS turns off at 37 percent, so in the UI it will remain at 37 percent, then the power will come back, and then suddenly it will jump to 0 percent when NUT reconnects to the UPS and new > info is available. In no case does it disappear or do anything else.
On the contrary, I would be happy if, instead of it disappearing from the list, it would simply turn it gray or something like that to indicate that it is unavailable.

I'll look into these and create sub-issues if necessary.