[Data] Add SERVICE_UNAVAILABLE to list of retried transient errors (#…

…47673)

While reading or writing files with Ray Data, S3 might raise a transient SERVICE_UNAVAILABLE error. This PR adds the error to the list of retried transient errors.

Signed-off-by: Balaji Veeramani <bveeramani@berkeley.edu>

python/ray/data/context.py

@@ -102,6 +102,8 @@
 DEFAULT_ENABLE_GET_OBJECT_LOCATIONS_FOR_METRICS = False
 
 
+# `write_file_retry_on_errors` is deprecated in favor of `retried_io_errors`. You
+# shouldn't need to modify `DEFAULT_WRITE_FILE_RETRY_ON_ERRORS`.
 DEFAULT_WRITE_FILE_RETRY_ON_ERRORS = (
     "AWS Error INTERNAL_FAILURE",
     "AWS Error NETWORK_CONNECTION",
@@ -114,6 +116,7 @@
     "AWS Error NETWORK_CONNECTION",
     "AWS Error SLOW_DOWN",
     "AWS Error UNKNOWN (HTTP status 503)",
+    "AWS Error SERVICE_UNAVAILABLE",
 )
 
 DEFAULT_WARN_ON_DRIVER_MEMORY_USAGE_BYTES = 2 * 1024 * 1024 * 1024