You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
S3 multipart upload now retries individual parts on transient transport errors (connection reset, timeout) with exponential backoff. Previously, a single transient failure on any part aborted the entire upload, wasting all successfully uploaded parts. This caused large file backups (~1 GB) to fail intermittently on constrained network paths such as Railway to Cloudflare R2.
Added a 300-second per-request timeout to S3 upload operations (previously only connect timeout was set).
Added
ONEIO_S3_MAX_RETRIES environment variable to configure retry attempts after the initial request for transient S3 transport errors (default: 3).
ONEIO_S3_RETRY_BACKOFF_MS environment variable to configure initial retry backoff in milliseconds (default: 1000, doubles each attempt).