Table of contents
1. The problem of file system power-fail reliability
1.1 Background – What exactly is a “file system”?
1.2 Possible failure modes
1.3 Mitigation tactics used by modern fi le systems
1.4 Use-case dependent severity
2. Systematic testing of fi le system power-fail immunity
2.1 Idea
2.2 Methodology
2.3 Completeness
3. Value for the customer
4. References
Introduction
“Everything is a file”, is one of the most often cited rules of thumb in the Linux ecosystem. Among other aspects, this illustrates how important files and thus reliability of file systems are for a Linux operating system. The easiest way to improve reliability is to configure the file system as read-only. This prohibits any modification of the content inside. Hence the file system cannot be harmed by failures during write operations. Of course, this is not always an option. Configuration data, log files, temporary information, and others need a file system that is reliable and writable.
Resource type
Tech paper
Related content
Click on the button to load the content from Pardot.
