Server Cache Data Capture During Asynchronous System Resets
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Information handling systems in datacenters often experience unexpected 'hangs' or 'crashes,' leading to permanent loss of uncommitted data due to system resets, which are difficult to identify and resolve quickly, especially when they result in asynchronous system resets that overwrite volatile memory.
Innovation Solution
A management resource, such as a service processor, maintains a flag to indicate the type of system reset, allowing the BIOS to copy server cache contents to persistent storage during an asynchronous reset, preserving data and enabling analysis of the crash origin.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a system reset is executed to resolve a hung system, then the system returns to a functional operational state, but uncommitted data in volatile memory is permanently lost
Solution Approach 1:
The patent applies preliminary action by capturing and storing system memory contents to persistent storage before the system reset permanently overwrites volatile memory. The management resource detects the reset condition and initiates a memory capture operation, preserving data in advance before it would be lost, thus resolving the contradiction between restoring system functionality and preventing data loss.
2Ease of operation
If system memory is overwritten during a reset to initialize the system, then the system can restart the operating system, but insight into the origin of the problem is lost
Solution Approach 1:
The patent applies copying by creating a duplicate copy of the system memory contents and storing it in persistent storage before the reset overwrites the original. This copy preserves diagnostic information about the crash origin while allowing the system to proceed with normal restart and initialization procedures, thus maintaining both system restart capability and diagnostic insight.
3Reliability
If a watchdog timeout event triggers an asynchronous system reset, then the hung system is resolved, but the reset type cannot be easily distinguished from a normal reset
Solution Approach 1:
The patent applies a metaphorical 'color change' by setting a persistent flag to a specific value (e.g., '1') that indicates an asynchronous reset occurred. This flag acts as a distinctive marker that differentiates asynchronous resets from normal synchronous resets, making reset type identification straightforward through simple flag checking, thus resolving the difficulty of detecting and measuring reset types.
Data Source
AI summary
A method and system respond to a system reset by executing a BIOS that establishes pre-OS communication with a baseboard management controller (BMC) and determines the state of a flag maintained by the BMC. The flag indicates whether the system reset was part of a crash event or a controlled shutdown. If the flag indicated a crash event, contents of the server cache are copied to a persistent secondary memory. The content of the secondary memory may then be sent to a management resource and/or to a remote management system. Before booting to an operating system and resuming normal operation, the flag may be SET. When a subsequent system resent occurs, the flag is cleared if the system reset is part of a controlled shutdown. Otherwise, the flag will remain set to indicate the crash event when the boot sequence is initiated.