From e83ae51d5fe3b53cef663334f0b4fcf75e2d3264 Mon Sep 17 00:00:00 2001
From: ilongin <ivan.longin1@gmail.com>
Date: Tue, 9 Jun 2026 00:35:13 +0200
Subject: [PATCH] Removing obsolete dataset statuses

---
 src/datachain/catalog/catalog.py        | 2 +-
 src/datachain/data_storage/metastore.py | 5 ++---
 src/datachain/dataset.py                | 3 ---
 3 files changed, 3 insertions(+), 7 deletions(-)

diff --git a/src/datachain/catalog/catalog.py b/src/datachain/catalog/catalog.py
index f3ce21d8c..7f0c8e471 100644
--- a/src/datachain/catalog/catalog.py
+++ b/src/datachain/catalog/catalog.py
@@ -1151,7 +1151,7 @@ def cleanup_dataset_versions(self, job_id: str | None = None) -> int:
         Clean up dataset versions that are no longer needed.
 
         Removes dataset versions that:
-        - Have status CREATED, FAILED, STALE, or REMOVING
+        - Have status CREATED, FAILED, or REMOVING
         - Belong to completed/failed/canceled jobs (not running)
         - Are session_* datasets from finished jobs (orphaned intermediates)
 
diff --git a/src/datachain/data_storage/metastore.py b/src/datachain/data_storage/metastore.py
index 1abd99007..f0b405ea6 100644
--- a/src/datachain/data_storage/metastore.py
+++ b/src/datachain/data_storage/metastore.py
@@ -343,7 +343,7 @@ def get_dataset_versions_to_clean(
         job (used during job failure cleanup).
 
         When job_id is None, returns all versions that are safe to delete:
-        - Status CREATED, FAILED, STALE where either:
+        - Status CREATED, FAILED where either:
           - the associated job has finished, or
           - there is no associated job (job_id is NULL) and the version is
             older than STALE_CREATED_THRESHOLD_HOURS
@@ -1831,13 +1831,12 @@ def get_dataset_versions_to_clean(
             )
             .where(
                 or_(
-                    # Incomplete/failed/stale versions from finished jobs
+                    # Incomplete/failed versions from finished jobs
                     and_(
                         dv.c.status.in_(
                             [
                                 DatasetStatus.CREATED,
                                 DatasetStatus.FAILED,
-                                DatasetStatus.STALE,
                                 DatasetStatus.REMOVING,
                             ]
                         ),
diff --git a/src/datachain/dataset.py b/src/datachain/dataset.py
index cd825d054..55aaa7de5 100644
--- a/src/datachain/dataset.py
+++ b/src/datachain/dataset.py
@@ -260,10 +260,8 @@ def __hash__(self):
 
 class DatasetStatus:
     CREATED = 1
-    PENDING = 2
     FAILED = 3
     COMPLETE = 4
-    STALE = 6
     REMOVING = 7
 
 
@@ -371,7 +369,6 @@ def is_final_status(self) -> bool:
         return self.status in [
             DatasetStatus.FAILED,
             DatasetStatus.COMPLETE,
-            DatasetStatus.STALE,
             DatasetStatus.REMOVING,
         ]