thds.adls 3.1.20250407135823__py3-none-any.whl → 3.1.20250408202524__py3-none-any.whl

thds/adls/download.py

@@ -5,6 +5,7 @@ import shutil
 import typing as ty
 from base64 import b64decode
 
+import aiohttp.http_exceptions
 from azure.core.exceptions import AzureError, HttpResponseError
 from azure.storage.filedatalake import (
     ContentSettings,
@@ -13,7 +14,7 @@ from azure.storage.filedatalake import (
     FileSystemClient,
 )
 
-from thds.core import log, scope, tmp
+from thds.core import fretry, log, scope, tmp
 from thds.core.hashing import b64
 from thds.core.types import StrOrPath
 
@@ -313,6 +314,13 @@ def _set_md5_if_missing(
     return file_properties.content_settings
 
 
+def _excs_to_retry() -> ty.Callable[[Exception], bool]:
+    """These are exceptions that we observe to be spurious failures worth retrying."""
+    return fretry.is_exc(
+        aiohttp.http_exceptions.ContentLengthError, aiohttp.client_exceptions.ClientPayloadError
+    )
+
+
 @_dl_scope.bound
 def download_or_use_verified(
     fs_client: FileSystemClient,
@@ -339,7 +347,10 @@ def download_or_use_verified(
                 co_request = co.send(file_properties)
             elif isinstance(co_request, azcopy.download.DownloadRequest):
                 # coroutine is requesting download
-                azcopy.download.sync_fastpath(dl_file_client, co_request)
+                fretry.retry_regular(_excs_to_retry(), fretry.n_times(2))(
+                    # retry n_times(2) means _retry_ twice.
+                    azcopy.download.sync_fastpath
+                )(dl_file_client, co_request)
                 co_request = co.send(None)
             else:
                 raise ValueError(f"Unexpected coroutine request: {co_request}")
@@ -378,7 +389,15 @@ async def async_download_or_use_verified(
                 co_request = co.send(file_properties)
             elif isinstance(co_request, azcopy.download.DownloadRequest):
                 # coroutine is requesting download
-                await azcopy.download.async_fastpath(dl_file_client, co_request)
+
+                await fretry.retry_regular_async(
+                    _excs_to_retry(), fretry.iter_to_async(fretry.n_times(2))
+                )(
+                    # retry n_times(2) means _retry_ twice.
+                    azcopy.download.async_fastpath
+                )(
+                    dl_file_client, co_request
+                )
                 co_request = co.send(None)
             else:
                 raise ValueError(f"Unexpected coroutine request: {co_request}")

thds/adls/download_lock.py

@@ -1,4 +1,4 @@
-import random
+import stat
 import time
 from datetime import timedelta
 from pathlib import Path
@@ -11,6 +11,8 @@ from .md5 import hex_md5_str
 
 DOWNLOAD_LOCKS_DIR = config.item("dir", home.HOMEDIR() / ".adls-md5-download-locks", parse=Path)
 _CLEAN_UP_LOCKFILES_AFTER_TIME = timedelta(hours=24)
+_CLEAN_UP_LOCKFILES_EVERY = timedelta(hours=1).total_seconds()
+_LAST_CLEANED_BY_THIS_PROCESS = time.monotonic() - _CLEAN_UP_LOCKFILES_EVERY
 logger = log.getLogger(__name__)
 
 
@@ -19,7 +21,8 @@ def _clean_download_locks() -> int:
     deletion_threshold = time.time() - _CLEAN_UP_LOCKFILES_AFTER_TIME.total_seconds()
     try:
         for f in DOWNLOAD_LOCKS_DIR().iterdir():
-            if f.is_file() and f.stat().st_mtime < deletion_threshold:
+            fstat = f.stat()
+            if stat.S_ISREG(fstat.st_mode) and fstat.st_mtime < deletion_threshold:
                 f.unlink()
                 deleted += 1
     except Exception:
@@ -31,8 +34,11 @@ def _clean_download_locks() -> int:
 
 
 def _occasionally_clean_download_locks():
-    if random.random() < 0.005:  # do this about every 200 downloads
-        # random.random is considered to be very fast, and we have no need of cryptographic quality.
+    global _LAST_CLEANED_BY_THIS_PROCESS
+    # do this about once an hour
+    if time.monotonic() > _LAST_CLEANED_BY_THIS_PROCESS + _CLEAN_UP_LOCKFILES_EVERY:
+        _LAST_CLEANED_BY_THIS_PROCESS = time.monotonic()
+        # minor race condition with other threads but it doesn't really matter.
         _clean_download_locks()
 
 

{thds_adls-3.1.20250407135823.dist-info → thds_adls-3.1.20250408202524.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: thds.adls
-Version: 3.1.20250407135823
+Version: 3.1.20250408202524
 Summary: ADLS tools
 Author-email: Trilliant Health <info@trillianthealth.com>
 License: MIT

{thds_adls-3.1.20250407135823.dist-info → thds_adls-3.1.20250408202524.dist-info}/RECORD

@@ -7,8 +7,8 @@ thds/adls/conf.py,sha256=q1SPrgb46NpobVzwt_Oyv71-BvsIbZLq9nRWS3LZjz0,1990
 thds/adls/copy.py,sha256=jUWbGvTpb4B3yRGS0nhGSbDzqRPzUqYgH0z1lFRJB3k,6365
 thds/adls/dbfs.py,sha256=pPAjbIZRKJsaXKQljDMUgqS_zy1yKeEZHGMueXbuv3g,2219
 thds/adls/defaults.py,sha256=GGq5Pn4r-8cX4bZItp4nnwWAAz7S07pzPoOegw0y5Fw,676
-thds/adls/download.py,sha256=JUF-J6bfaSzDM389errg8tHPssGt-K1yp3O_pznls3o,16852
-thds/adls/download_lock.py,sha256=_JZj-kjCUfHk9FvrmEuYpJYknmbam5eReFhGNDgzdLQ,2520
+thds/adls/download.py,sha256=ds8makvLOiC8LibHzb_X85XLaSPGMCNxG7ZL7bEVzCI,17584
+thds/adls/download_lock.py,sha256=ttD2GhPNRnITNoV1XH2PvKbMsHppZirjy3RZ4P4kgKM,2826
 thds/adls/errors.py,sha256=B_rMsQvQnNmP_sf-x8kmGsv2vIeOh4G9kVbdNVyk350,1469
 thds/adls/etag.py,sha256=ct7jpHhNFcKzbekn5rZ3m6DhjK48A7qOZGwDiHkc-pc,242
 thds/adls/file_properties.py,sha256=V3VEjEG3PNyeQaJ159Kco5l7c2EiyjE0yygtmVVjd6E,1597
@@ -33,8 +33,8 @@ thds/adls/resource/up_down.py,sha256=3uNlTvm2gVhSyYdQTBwsGecOgwtINQfINckR-awwV0Y
 thds/adls/tools/download.py,sha256=vvBO8lSDl9oPugv75qpCkoemT9pOM9BV6yeExlkyG08,1594
 thds/adls/tools/ls.py,sha256=OgEaIfTK359twlZIj-A0AW_nv81Z6zi0b9Tw6OJJfWA,1083
 thds/adls/tools/upload.py,sha256=eMk4pdug1aCMPDDWpIE3Zoq77i5APp9Uuh-sVCCDNJE,493
-thds_adls-3.1.20250407135823.dist-info/METADATA,sha256=uMWMLqoD4fkF67t6XXhyJ4h5YgaNCWHJlxJmCJIWATc,548
-thds_adls-3.1.20250407135823.dist-info/WHEEL,sha256=iAkIy5fosb7FzIOwONchHf19Qu7_1wCWyFNR5gu9nU0,91
-thds_adls-3.1.20250407135823.dist-info/entry_points.txt,sha256=uTqreT1AIwqJboMfLv5w6sviM8mNbAkln765gIjzoA4,152
-thds_adls-3.1.20250407135823.dist-info/top_level.txt,sha256=LTZaE5SkWJwv9bwOlMbIhiS-JWQEEIcjVYnJrt-CriY,5
-thds_adls-3.1.20250407135823.dist-info/RECORD,,
+thds_adls-3.1.20250408202524.dist-info/METADATA,sha256=uN4O6WEdsKfOuz340C5XnDYiuQWvWQcDnY9f5SRrBRY,548
+thds_adls-3.1.20250408202524.dist-info/WHEEL,sha256=iAkIy5fosb7FzIOwONchHf19Qu7_1wCWyFNR5gu9nU0,91
+thds_adls-3.1.20250408202524.dist-info/entry_points.txt,sha256=uTqreT1AIwqJboMfLv5w6sviM8mNbAkln765gIjzoA4,152
+thds_adls-3.1.20250408202524.dist-info/top_level.txt,sha256=LTZaE5SkWJwv9bwOlMbIhiS-JWQEEIcjVYnJrt-CriY,5
+thds_adls-3.1.20250408202524.dist-info/RECORD,,