tomholford f22a1b05a8 protondrive: implement shouldRetry instead of always returning false ...

shouldRetry was a stub returning false unconditionally, which makes
protondrive the only rclone backend that disables pacer-level retries
entirely. Every other backend at minimum falls back to
fserrors.ShouldRetry(err) so genuine transport-level transients (TCP
resets, brief 5xx) get retried.

- Use errors.As to unwrap proton.APIError instead of string matching
- Retry transient storage block errors (Code=200501)
- Retry server errors (5xx, except 503)
- Skip 429 and 503 (handled by go-proton-api's resty retry layer
  via catchTooManyRequests / catchRetryAfter, which honours Retry-After)
- Fall back to fserrors.ShouldRetry for non-API errors

Co-authored-by: tomholford <tomholford@users.noreply.github.com>

2026-05-24 17:50:01 +01:00

..

protondrive_internal_test.go

protondrive: implement shouldRetry instead of always returning false

2026-05-24 17:50:01 +01:00

protondrive_test.go

protondrive: add protondrive backend - fixes #6072

2023-07-22 10:46:21 +01:00

protondrive.go

protondrive: implement shouldRetry instead of always returning false

2026-05-24 17:50:01 +01:00