Server Coolant Leak Detection Using Node Voltage Fault Alerts
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for detecting coolant leaks in server systems are inefficient, often leading to significant leaks going undetected until a drip pan is filled, causing potential damage to servers due to voltage faults.
Innovation Solution
A controller system that receives voltage fault alerts from multiple nodes in a server system and powers down nodes below a detected leak to prevent further damage, while also triggering a shut-off valve to stop the leak.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a drip pan with a sensor is used to detect coolant leaks, then the system can detect leaks, but by the time detection occurs a significant leak has already happened causing potential server damage
Solution Approach 1:
The patent applies preliminary action by monitoring voltage faults on individual server nodes before a significant leak occurs. The system proactively detects early signs of coolant leakage through voltage anomalies and triggers shutdown sequences in advance, preventing the leak from reaching the drip pan sensor threshold and causing server damage.
Solution Approach 2:
The patent replaces the mechanical drip pan sensor system with an electrical monitoring system that detects voltage faults on server nodes. This substitution enables earlier detection by monitoring electrical parameters rather than waiting for mechanical accumulation of coolant in the drip pan.
2Reliability
If a rope sensor is used to detect coolant leaks, then the system can detect leaks, but it is difficult to snake the rope around each computer in the chassis so leaks may be missed
Solution Approach 1:
The patent applies self-service by leveraging the existing voltage monitoring infrastructure already present on each server node. Each node's power supply unit (PSU) inherently monitors voltage, and the system utilizes these existing sensors without requiring additional installation of ropes or external monitoring devices around each chassis component.
Solution Approach 2:
The patent applies universality by using the existing voltage monitoring system, originally designed for power management, to serve the additional function of coolant leak detection. This multi-functional use of existing infrastructure eliminates the need for dedicated leak detection sensors in each node.
3Loss of time
If voltage fault alerts from multiple nodes are monitored, then the system can detect leaks more quickly and accurately, but the system complexity increases
Solution Approach 1:
The patent applies segmentation by monitoring voltage faults on individual server nodes independently rather than using a single centralized sensor. Each node is monitored separately, allowing the system to identify the specific location and progression of a leak through sequential voltage fault alerts from different nodes.
Solution Approach 2:
The patent applies feedback by continuously monitoring voltage fault alerts from multiple nodes and using this information to dynamically control the shutdown sequence. The controller receives feedback from each node's voltage status and adjusts the shutdown progression accordingly, powering down nodes sequentially from the one with the earliest voltage fault downward.
Data Source
AI summary
Detecting coolant leaks in a server system including receiving, by a controller in the server system with nodes stacked vertically, a voltage fault alert from a first node; receiving, by the controller in the server system with nodes stacked vertically, a voltage fault alert from a second node located below the first node; and upon receiving the voltage fault alert from the first node and the voltage fault alert from the second node, powering down, by the controller, one or more nodes located below the second node.