sc_analysis_dump

Output the traceroute of warts files one after the other in an easy-to-parse format

$ sc_analysis_dump [path/to/file1.warts path/to/file2.warts ...]

sc_analysis_dump [-d <directory> | -o <file>] [-t <startTime>:<endTime>] [-x <dumpOption>] [-f] [-v]

-d <directory>
    Specifies the directory where `sc_analysis` collected the analysis data to be dumped from. This is often an internal data repository.

-o <file>
    Specifies the output file where the dumped analysis data will be written. Note that typically only one of `-d` (source) or `-o` (destination) is used to define the primary operation context.

-t <startTime>:<endTime>
    Defines a specific time range for the data to be dumped. Times are usually specified in Unix epoch format or a specific system-defined timestamp format like YYYYMMDD.HHMMSS.

-x <dumpOption>
    Specifies the type of data to dump. Common options include all (for comprehensive data), summary, io, latency, stats, events, fileset, or inode, allowing for targeted data extraction based on diagnostic needs.

-f
    Forces the dump operation, potentially overriding warnings or existing files without prompting for confirmation.

-v
    Enables verbose output, providing more detailed information about the dump process, which can be useful for debugging the command itself.

The `sc_analysis_dump` command is a utility within the IBM Spectrum Scale (formerly GPFS) file system suite. Its primary function is to extract and dump previously collected performance and diagnostic data, which was gathered by the `sc_analysis` command. This data is crucial for in-depth problem determination, performance analysis, and often used by IBM support for advanced diagnostics. It allows administrators to retrieve specific datasets, often filtered by time range or data type, to investigate system behavior, bottlenecks, or anomalies within the Spectrum Scale cluster. The output is typically in a specialized format designed for further processing by IBM's internal analysis tools, rather than direct human readability. It is an essential tool for deep-dive investigations into Spectrum Scale cluster health and performance.

This command is specific to IBM Spectrum Scale (GPFS) environments and is not a standard Linux utility found on generic distributions. It critically requires that analysis data has already been actively collected and stored by the `sc_analysis` command or daemon. The output format is highly specialized, typically machine-readable for IBM's diagnostic tools, and generally not intended for direct human consumption or parsing with standard text utilities. Running this command on large datasets or for extended time ranges can generate extremely large output files, potentially consuming significant disk space.

The data generated by `sc_analysis_dump` is often in a highly optimized, internal binary or structured format (e.g., Protobuf or similar) specifically designed for consumption by IBM's proprietary analysis tools and scripts. It is generally not intended for direct human interpretation or parsing with standard text processing tools like `grep` or `awk`.

For `sc_analysis_dump` to function correctly, the system must be an IBM Spectrum Scale cluster node, and crucially, analysis data must have been actively collected and stored by the `sc_analysis` daemon or process prior to running this command. Without pre-collected data, the command will have nothing to dump.

The `sc_analysis_dump` command has been an integral part of the diagnostic toolkit for IBM General Parallel File System (GPFS), which was later rebranded as IBM Spectrum Scale. It was developed by IBM to support its high-performance, distributed file system, enabling administrators and support staff to deeply analyze system behavior. It has evolved alongside the file system, adapting to new data types and performance metrics introduced in successive Spectrum Scale versions, ensuring comprehensive analysis capabilities for complex enterprise storage environments.

sc_analysis(1), mmdiag(1), mmfsenv(1), mmlscluster(1), mmstat(1)