gedih3.doctor.diagnoses.soc_health#

soc_health diagnosis — invalid HDF5 files + download log drift.

Detects:

  • SOC HDF5 files that fail h5_is_valid().

  • Download log entries marked DOWNLOADED but with the file missing or empty.

Remedies (safe):

  • Mark drifted entries FAILED so the next gh3_download --resume retries.

  • Invalid HDF5 files are reported only (could be in-progress); deletion is not auto-applied (see --delete-invalid future flag).

Performance pillar (v0.8.x lessons backport):

  • SOC file enumeration always walks the tree in parallel via gedih3.parallel.walk_soc_parallel(). The legacy _soc_manifest.txt shortcut was removed from the read path because external population (manual rsync, NASA delivery) bypasses the producer-driven refresh and silently narrows the doctor’s view.

  • Per-file h5_is_valid checks are dispatched in parallel via gedih3.doctor.parallel.parallel_map() when a dask client is registered.

Functions#

soc_health_check(→ gedih3.doctor.report.Report)

soc_health_fix(→ gedih3.doctor.report.Report)

Module Contents#

gedih3.doctor.diagnoses.soc_health.soc_health_check(ctx: gedih3.doctor.report.DoctorContext) → gedih3.doctor.report.Report[source]#
gedih3.doctor.diagnoses.soc_health.soc_health_fix(ctx: gedih3.doctor.report.DoctorContext, report: gedih3.doctor.report.Report) → gedih3.doctor.report.Report[source]#