Power calculations show Allocated=0 with power rails and modules

[markkuleinio]

I'm trying to document the power connections and usage in the rack. This is the physical world:

• Datacenter has two independent power systems
• Each rack has two feeds: Feed1 and Feed2 (connected to different upstream power systems)
• Each rack has one Rittal power rail (actually "PSM busbar" but it is "rail" in this post...):
  • The rail has two upstream cables (both are 3 x 230V x 16A): one cable for Feed1, the other for Feed2
  • The rail has seven slots:
    • Each slot accepts a one-phase module like a "6 x C13" module (or various others, like "4 x schuko" etc)
    • The slots are in different phases: 1, 2, 3, 1, 2, 3, 1 (= slots 1, 4 and 7 are phase 1, and so on)
    • The orientation of the module in the slot determines the feed to be used: if the "red arrow" in the module is up, it is using Feed1, if the arrow is down, then it is using Feed2

(I believe the system is Rittal Power System Module PSM)

One detail to note that there are no circuit breakers in the power rails or in the modules (other than maybe some temperature-activated ones), so the "max = 16A" is determined by the circuit breaker in the power feeds.

Now to exact NetBox modeling. I have:

• two Power Panels (Panel1 and Panel2)
• two Power Feeds (like "Rack1 Feed1" = AC, 400V, 16A, 3-phase)
• one power rail "Rack1 Rail": this is a Device with:
  • two Power Ports "Feed 1" and "Feed 2" (one connected to "Rack1 Feed1", the other connected to "Rack1 Feed2")
  • seven Power Outlets "Slot[1-7]" (for purpose of this demo, I have Slot3 configured: power port = Feed 1, feed leg = C)
• one power module "Rack1 Slot3": this is a Device with:
  • one Power Port "Backplane" (connected to "Rack1 Rail" port Slot3)
  • six Power Outlets "[1-6]" (port 1 configured: power port = Backplane, feed leg = C)
• one power-consuming device "Nexus1", with:
  • two Power Ports "PS1" and "PS2", with max draw = 500W, allocated draw = 120W
  • PS1 is connected to "Rack1 Slot3" port 1

Now, if I look at the power module "Rack1 Slot3", it says in the Power Utilization: Allocated = 120VA, which is correct.

Then I go to the upstream device "Rack1 Rail", but in Power Utilization there is Allocated = 0VA, and thus when I go to the feed "Rack1 Feed1", it says Utilization (Allocated) = 0VA/8868VA, which is incorrect.

Now the question is: Am I doing something incorrectly? I'm expecting to see the Allocated power figures in the upper levels as well, to get the actual calculations up to the circuit breakers.

Bonus: This is currently configured in demo.netbox.dev (start for example from https://demo.netbox.dev/dcim/power-feeds/49/) before it gets purged in the night or so. (Disclaimer: someone is already testing something else with these configs so it's not my original setup anymore.)

[markkuleinio]

One observation is that the Power Feed only reports one utilization value. Maybe it is assumed that a three-phase feed is only used for cases where all three phases are utilized equally (like three-phase motors I'd assume; if you run large motors in your datacenter 😄).

In my example though the three phases are utilized independently, so even if the values were inherited correctly to the Power Feed level, that wouldn't give me any meaningful value.

So, maybe I need to create six separate (single-phase) feeds ("Rack1 Feed1A", "Rack1 Feed1B", "Rack1 Feed1C", ...) for one rack to report total usage correctly. (But this is irrelevant to the original calculation problem AFAIK).

[markkuleinio]

As someone already tried, if "Rack1 Slot3" power port is manually configured to an Allocated draw value (like 1000W), then "Rack1 Feed1" shows the utilization value (1000).

So it looks like there is a limit how far the configured Allocated draw values will traverse upstream.

For generic cases I could leave out the power modules and just configure the power rail directly with all the Power Outlets, then the utilization should be correct in the feed. But, it falls apart again when I need to model an extra power strip (a device that consumes 1U in the rack so it's very much interesting device) connected to the power rail.

[markkuleinio]

As a reference, #9148 has been opened in relation to this.