fluid diagnose collects diagnostic artifacts for a Fluid Dataset: CR YAML, pod descriptions, logs, namespace events, PVC/PV data, and an optional tar.gz archive for support.
- Opening a GitHub issue or vendor support ticket
- Capturing cluster state after an incident
- Sharing a timestamped bundle with your team
Use fluid inspect for a lightweight resource listing without log collection.
# Collect into a timestamped directory (default TUI to browse results)
fluid diagnose my-dataset -n default
# Write artifacts only (no TUI)
fluid diagnose my-dataset -n default -o dir
# Create a tar.gz archive for attachment
fluid diagnose my-dataset -n default --archive -o dirArtifacts are written under a directory such as fluid-diagnose-<dataset>-<timestamp>/:
| Path | Contents |
|---|---|
dataset.yaml |
Dataset CR |
dataset.describe.txt |
Human-readable Dataset summary |
runtime/ |
Runtime CRs and descriptions |
pods/ |
Per-pod YAML, describe, and logs |
events/ |
Namespace events |
storage/ |
PVC and PV YAML |
controllers/ |
Fluid controller logs (if --include-controller-logs) |
summary.txt |
High-level summary and warnings |
manifest.json |
Index of every artifact and collection status |
manifest.json records each file with status (collected, failed, skipped) and an optional reason. Partial failures are normal when pods are missing or logs are unavailable; check summary.txt and the manifest.
With --archive, a .tar.gz is produced alongside the directory (see command output for the path).
- Run diagnose with
--archiveand-o dirin non-interactive environments. - Redact secrets from the bundle if needed before sharing.
- Attach the archive to your issue along with
fluid versionoutput.
fluid diagnose --helpNotable options include --output-dir, --no-logs, --include-controller-logs, and --since (limit log/event age). Defaults and full descriptions are in --help.