Unsolved
1 Rookie
•
1 Message
•
2 Points
0
126
February 27th, 2026 03:10
Poweredge calculate power for Redundant policy
Hi all.
I am trying to clearly understand how to correctly design and size the PSU configuration on a PowerEdge R760 in order to reliably enable and maintain Grid A/B Redundancy.
Current scenario:
Server model: PowerEdge R760
2 × 800W identical PSUs (same part number, same firmware)
Peak observed system consumption ~465W (via iDRAC monitoring)
Both PSUs show Healthy status
Grid A/B Redundancy policy enabled
Both PSUs connected to the same UPS
The system reports “PSU Redundancy Lost.”
I would like clarification on the following points:
1. Grid A/B power source requirements
Does iDRAC detect whether both PSUs are connected to the same upstream source (same UPS / PDU / phase)?
Is Grid A/B strictly requiring two electrically independent feeds?
If both PSUs are connected to the same UPS, will the system always report Redundancy Lost?
Official documentation states that Grid A/B implies that each PSU must be able to support 100% of system load, but it does not clearly explain how upstream power source independence is evaluated.
2. How to properly calculate PSU sizing for guaranteed redundancy
For example:
If peak consumption is ~465W,
And each PSU is rated 800W,
This should theoretically satisfy 1+1 redundancy.
However:
Does the system use observed peak,
Calculated worst-case configuration power,
Or some internal power budgeting algorithm?
How can we deterministically calculate PSU wattage to guarantee Grid A/B will be considered valid?
Is there an official sizing formula or guideline for 16G platforms?
3. Where can we see the internal power calculation basis?
Is there a way in iDRAC to view:
Maximum calculated system power requirement,
Power budget used for redundancy validation,
Any internal “worst-case load” estimate,
The specific reason why redundancy is considered lost?
For example:
Does iDRAC evaluate component TDP (CPU, memory, PCIe, GPUs)?
Does it use PSU derating rules?
Is there a log entry that explains the exact failure condition?
4. Documentation request
Could you please provide:
Official documentation describing Grid A/B validation logic for 16G servers,
PSU sizing best practices for guaranteed redundancy,
Any relevant technical whitepaper explaining redundancy evaluation logic?
We need deterministic criteria to properly design data center power layout and avoid ambiguity when deploying R760 servers in production environments.
Thank you in advance for clarification.
0 events found


Dell-Martin S
Moderator
•
3.6K Posts
•
16 Points
0
March 2nd, 2026 07:28
Hi,
iDRAC treats Grid A/B as purely a PSU‑capacity/topic‑of‑failure concept, not as an automatic detector of whether your PSUs are fed from independent UPS/PDUs/ phases.[1][2][3]
## 1. Grid A/B power source requirements
- iDRAC does **not** detect whether both PSUs are on the same UPS, PDU, or electrical phase; it only knows if each PSU has valid input, output, and status.[1][3]
- A/B Grid Redundant means: PSUs are split into grid A and grid B (in a 2‑PSU R760 that is effectively N+N), and the platform power budget must be such that one whole grid can carry the entire calculated worst‑case load.[1][2][3]
- The requirement for “two electrically independent feeds” is a **design best practice**, not something iDRAC validates; if you plug both PSUs into the same UPS, Grid A/B can still be reported as healthy if the capacity/budget rules are satisfied.[1][2][4][3]
- iDRAC will only show “PSU Redundancy Lost” when, according to its internal power model and PSU ratings/conditions, one grid by itself cannot support the full calculated requirement, or when one PSU/grid is faulted or offline.[1][5][3]
So: having both PSUs on the same UPS does **not** by itself force “Redundancy Lost”; the alert is driven by calculated capacity or PSU health, not feed independence.[1][3]
## 2. How PSU sizing is actually evaluated
We don't use your observed peak (e.g. 465 W) to decide redundancy state; Grid A/B / PSU Redundant are based on an internal power budget for the *configured* system, plus PSU capabilities and derating, not on telemetry history.[2][6][3]
Key points:
- The platform uses a configuration‑based maximum power requirement that takes into account CPUs, memory population, NVMe/SAS drives, PCIe cards, GPUs/accelerators, fans, etc.[2][6][3]
- This “worst‑case configuration power” is compared against the *available power per grid* (one PSU or set of PSUs) after considering:
- PSU nameplate rating and input voltage (de‑rated at low line),
- thermal/efficiency derating region,
- redundancy policy (A/B grid vs PSU redundant vs not redundant).[2][6][3]
- If $$ \text{available power of one grid} < \text{calculated maximum configuration power} $$, redundancy is flagged as lost, even if your real‑world load is far lower (e.g. 465 W).[2][6][3]
For a 2 × 800 W R760 with Grid A/B:
- On high‑line input (e.g. 200–240 V), each 800 W PSU can typically deliver the full 800 W; on low line (100–120 V), its maximum usable output may be lower, which is taken into account for redundancy evaluation.[2][3]
- If the internal model for your particular CPU/MEM/PCIe/GPU configuration says “this system may require >800 W worst‑case,” iDRAC will show “PSU Redundancy Lost” under Grid A/B, even though iDRAC telemetry shows ~465 W peak.[2][6][3]
Deterministic sizing for 16G:
- The recommended deterministic method is to use the Enterprise Infrastructure Planning Tool (EIPT) / Online Power Calculator and the **PSU Guide** for your platform, and then choose the smallest PSU option marked as “Full Redundancy” (FR) for that configuration.[6][3]
- In EIPT, PSU capacities are color‑coded:
- Gray – cannot support configuration;
- White – Fault Tolerant Redundancy (supported but may throttle/degrade after redundancy loss);
- Green – Full Redundancy (minimum PSU capacity that supports full performance before and after redundancy loss).[6]
So to “guarantee” Grid A/B is considered valid:
1. Model your exact R760 configuration in EIPT.
2. Select PSU wattage where the tool marks the configuration as FR/green with N+N (A/B) redundancy.
3. Ensure input voltage matches that assumption (e.g. high‑line vs low‑line).
With that, the same sizing logic used by iDRAC for redundancy validation is satisfied.[2][6][3]
## 3. Where to see the internal power basis / reasons
You cannot see the exact full formula, but you can get several key pieces:
- iDRAC exposes:
- Current and historical power consumption and its own “Power Cap” limits,
- PSU ratings, input type (high‑line/low‑line), and status,
- Redundancy mode (Not Redundant / A/B Grid / PSU Redundant).[3]
- The **calculated maximum configuration power** and redundancy validation are **not** shown as a direct number in iDRAC, but are embedded in:
- EIPT’s “Estimated Maximum” / “Full Redundancy required PSU size” output for your entered configuration,
- BIOS/iDRAC power management policy behavior and alerts.[2][6][3]
What the internal model considers (per Dell power‑management whitepapers):
- Component‑level TDP and configuration rules (CPU SKUs, memory DIMM count/type/speed, NVMe/SAS drive count, PCIe / GPU type and count).[2][6]
- PSU derating based on input voltage and temperature; “maximum usable capacity” can be less than nameplate rating at low line or high ambient.[2][6][3]
- Platform overhead for fans, chipset, NICs, etc.[2][6]
How to see reasons/alerts:
- System Event Log (SEL) / iDRAC “Hardware -> Power -> Events” will show entries like `RDU0012 Power supply redundancy is lost` when redundancy evaluation fails; these do not include the numeric power comparison but are your main indicator.[5][3]
- There is no per‑component breakdown of the internal “worst‑case” model in GUI/CLI; that detail is abstracted into EIPT and design documentation.[2][6][3]
## 4. Documentation and best‑practice references
Key official materials for 16G PSU redundancy and sizing:
- “PowerEdge: Power Settings” knowledge base article – explains A/B Grid Redundant, PSU Redundant, and how redundancy depends on configuration and PSU capacity.[3]
- “Strategically Manage Power in PowerEdge Server PSUs Using …” whitepaper – describes redundancy modes (N+N A/B grid, N+1, non‑redundant), derating, and how PSUs are sized against platform config.[2]
- “Full Redundancy vs. Fault Tolerant Redundancy for PowerEdge Server PSUs” – discusses how 15G/16G tools classify PSU options (FR vs FTR) for a given configuration and shows the logic for required PSU capacity.[6]
- Power distribution / data center design whitepapers (e.g. chassis/PDUs) – describe best practices for mapping PSUs to separate PDUs/circuits to achieve *true* grid independence.[4]
For your specific R760 deployment:
- Use EIPT with your exact R760 BOM to obtain “Estimated Maximum” and the minimum PSU capacity flagged as FR in N+N mode, and choose that or higher.
- Ensure the server’s input voltage and environment match the assumptions in the calculator.
- Wire PSUs to independent PDUs/UPS feeds to achieve *actual* grid redundancy, even though iDRAC does not enforce or detect feed independence.[2][4][6][3]
Sources:
[1] PowerEdge PSU redundancy options - differences? https://www.dell.com/community/en/conversations/poweredge-hardware-general/poweredge-psu-redundancy-options-differences/647fa024f4ccf8a8de51abad
[2] [PDF] Strategically Manage Power in PowerEdge Server PSUs Using ... https://objects.icecat.biz/objects/mmo_135024374_1758638246_2314_703856.pdf
[3] PowerEdge: Power Settings | Dell US https://www.dell.com/support/kbdoc/en-us/000202926/poweredge-power-settings
[4] Power Distribution Systems for the ... https://i.dell.com/sites/csdocuments/Business_smb_merchandizing_Documents/pt/br/pe_m1000e_selection_whitepaper.pdf
[5] PowerEdge: PSU Redundancy Lost - part 5G4WK | Dell US https://www.dell.com/support/kbdoc/en-us/000215889/psu-redundancy-lost-dell-dpn-5g4wk
[6] Full Redundancy vs. Fault Tolerant ... https://infohub.delltechnologies.com/en-us/p/full-redundancy-vs-fault-tolerant-redundancy-for-poweredge-server-psus/
[7] Stopping iDRAC Reporting "Power supply redundancy is lost" as I ... https://www.dell.com/community/en/conversations/power-cooling/stopping-idrac-reporting-power-supply-redundancy-is-lost-as-i-have-moved-from-2-psus-to-one/685040b816c3a94aace0c4e9
[8] PSU Failover behaviour in r720 https://www.reddit.com/r/homelab/comments/u70tll/psu_failover_behaviour_in_r720/
[9] Power Supply troubleshooting for the ATP or SEDR 8880 and ... https://knowledge.broadcom.com/external/article/174386/power-supply-troubleshooting-for-th
e-atp.html
[10] Stop R710 iDrac complaining about loss of power to spare PSU https://www.reddit.com/r/homelab/comments/eaj7e4/stop_r710_idrac_complaining_about_loss_of_power/