pudl.scripts.dghome

Manage UUID-named directories in a Dagster storage directory.

This CLI helps manage the Dagster storage directory to reclaim disk space by listing and removing run artifacts (represented as UUID-named directories) based on modification date.

Examples

List all UUID-named directories:

dghome ls

List directories modified on or before 2026-01-15:

dghome ls 2026-01-15

List directories modified 10 days or more ago:

dghome ls 10d

Remove directories modified 1 month or more ago:

dghome rm 1m

Attributes

UUID_PATTERN
_DATE_FORMATS_EPILOG

Functions

_parse_date(→ datetime.datetime)	Parse a cutoff datetime from one of several formats.
_human_readable(→ str)	Format a kilobyte count as a human-readable string (e.g. 1.2M, 3.4G).
_row_color(→ str | tuple[int, int, int])	Get the right color of output for the filesize.
_collect(→ list[dict])	Return UUID dirs from $DAGSTER_HOME/storage, filtered to mtime <= cutoff_ts.
dghome(→ None)	Manage UUID-named directories in Dagster storage.
ls(→ None)	List UUID directories, optionally filtered by modification date.
rm(→ None)	Remove UUID directories last modified on or before DATE.
main(→ None)	Entry point for the dghome CLI.

Module Contents

pudl.scripts.dghome.UUID_PATTERN

pudl.scripts.dghome._parse_date(date_str: str, now: datetime.datetime | None = None) → datetime.datetime

Parse a cutoff datetime from one of several formats.

Accepts any absolute time string that dateutil can handle, plus relative time strings like “<N>d/w/m” (for days/weeks/months).

pudl.scripts.dghome._human_readable(kb: int) → str

Format a kilobyte count as a human-readable string (e.g. 1.2M, 3.4G).

pudl.scripts.dghome._row_color(size_kb: int) → str | tuple[int, int, int]

Get the right color of output for the filesize.

< 100 MB: green < 1 GB: yellow < 5 GB: orange else: red

pudl.scripts.dghome._collect(cutoff_ts: float | None) → list[dict]

Return UUID dirs from $DAGSTER_HOME/storage, filtered to mtime <= cutoff_ts.

Raises click.ClickException if DAGSTER_HOME is unset or the storage dir is missing. Results are sorted by modification time (oldest first).

pudl.scripts.dghome.dghome() → None

Manage UUID-named directories in Dagster storage.

pudl.scripts.dghome._DATE_FORMATS_EPILOG = Multiline-String

Show Value

"""
Accepts any absolute time string that dateutil can handle, plus relative time strings like '<N>d/w/m' (for days/weeks/months)."""

pudl.scripts.dghome.ls(date: str | None) → None

List UUID directories, optionally filtered by modification date.

Without DATE, lists all directories. With DATE, lists only those last modified on or before DATE.

pudl.scripts.dghome.rm(date: str | None) → None

Remove UUID directories last modified on or before DATE.

Without DATE, reports what would be removed but does nothing.

Confirms before deletion.

pudl.scripts.dghome.main() → None

Entry point for the dghome CLI.