Single Pass File System Repair Using Copy On Write
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional file system checkers require two passes through the file system, leading to unacceptable downtime and potential data corruption due to synchronization issues between memory and disk, and can introduce artifacts or false errors during the repair process.
Innovation Solution
Implementing a single pass file system repair using a copy on write approach, where repairs are written to a buffer instead of the active file system, allowing for immediate reflection of fixes and avoiding immediate changes to the file system until user confirmation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a two pass approach is used to check and repair file system errors, then repair completeness is improved, but file system downtime increases
Solution Approach 1:
The patent applies preliminary action by performing the read-only detection pass first to identify all errors before making any write operations. This allows the system to plan the complete repair sequence in advance, then execute all repairs in a single subsequent pass while the file system is mounted read-write, thereby reducing total downtime while maintaining repair completeness
Solution Approach 2:
The patent segments the file system checking process into distinct phases: a read-only detection phase that identifies errors without making repairs, and a read-write repair phase that executes fixes. This segmentation allows error detection to be completed first, enabling optimized repair scheduling that minimizes file system unavailability
2Productivity
If repairs are made during the first pass, then productivity is improved, but measurement precision deteriorates due to artifacts
Solution Approach 1:
The patent performs all error detection in a preliminary read-only pass before making any repairs. This ensures that the error inventory is complete and accurate before repair operations begin, eliminating artifacts caused by sequential detection and repair. Only after the complete error set is identified does the system proceed to the repair phase
3Device complexity
If sequential repair of errors is performed, then device complexity is reduced, but reliability worsens due to new errors
Solution Approach 1:
The patent performs preliminary detection of all errors in a read-only pass before executing any repairs. This allows the complete set of errors to be identified and scheduled for repair in advance, ensuring that repairs are made in an optimized sequence that prevents new errors from being introduced while maintaining system simplicity
Solution Approach 2:
The system uses feedback from the read-only detection pass to inform the repair strategy. By knowing all errors in advance, the system can adjust its repair sequence to avoid operations that might create new inconsistencies, thereby maintaining data consistency while keeping the overall system design simple
Data Source
AI summary
Example apparatus and methods concern identifying an error in a file system. The error is identified during one or more b-tree traversals through the file system. A fix for the error is produced upon detecting the error. Data or metadata associated with the fix is not initially written to the file system, but instead is stored somewhere other than the file system using a copy on write approach. After the traversals are complete, the file system may be fixed using the data or metadata stored using the copy on write approach.