ArgosDoes data survive faults?
Exercise interrupted operations, degraded reads, reconstruction, and consistency checks. Separate a successful recovery from an untested failure mode.
Research & reproducibility
Follow the evidence.
Reproduce the run.
ArgosFS provides a testbed for placement, recovery, and storage automation. Experiment scripts retain raw records and summaries so behavior can be inspected.
Questions to investigate
What should a storage
controller earn your trust on?
Does it adapt usefully?
Study workload shifts and uneven disks. Measure placement behavior and maintenance cost alongside foreground I/O.
Can decisions be explained?
Compare manual and autopilot policies. Inspect selected actions, rejection reasons, and verification outcomes instead of relying on a single final score.
Artifact workflow
Commands.
Records.
Results.
The experiment runner builds the release binary, runs its experiment families, and summarizes the output. Start with the quick mode.
Read the artifact guidescripts/experiments/run_all.sh --quick \
--output target/argosfs-artifacts/runs/local-quick“Quick” still builds Rust and may require extra tools or privileges for individual suites. Inspect failed and skipped records before interpreting summaries.
raw/JSONL and CSV experiment records
processed/ · tables/Summaries derived from the raw output
logs/ · manifest.jsonBuild output, commit, kernel, and seed
Experiment entry points
Pick a question. Inspect its script.
These are available experiment families, not published performance results. Hardware, mode, revision, and skipped checks are part of any comparison.
Contribute
Make the next result
easier to reproduce.
Report the revision, environment, command, and relevant logs with an issue.