Pause and cancel should not mean the same thing. In Flux, a paused task retains its saved transfer state so the existing task can continue. The important distinction is between bytes already stored on disk and bytes that a source allows the client to request again. Those two values do not always behave identically.
Resume the existing task
Select the paused task and use Resume. Adding the same URL as a new task creates a separate download rather than continuing the old checkpoint. Keep the staging data and destination available. Moving or deleting the task’s working files can force integrity checks or prevent a straightforward continuation.
When the server supports byte ranges
HTTP range requests let a client ask for a particular part of a file. Flux checks returned ranges and verifies stored checkpoints before combining old and new bytes. A strong ETag is not the only possible path to resume: the 2.5 engine can verify the saved prefix when that validator is absent.
When the server cannot resume directly
For a non-range source, Flux may replay the saved prefix from the network, verify it and discard the repeated bytes before appending new data. Your saved progress can remain while the connection still transfers that prefix again. This costs bandwidth and time; no download manager can make a server return a byte range it does not support.
If the remote file changed, continuing blindly could produce a mixed file. A validation failure is therefore different from a simple temporary network interruption. Review the task’s error and source before deciding to start over.
A reliable pause test
- Start a stable file and wait for nonzero disk progress.
- Pause the task and note the downloaded amount.
- Resume that task and watch its state and network activity.
- If the source changed or expired, investigate that cause instead of repeatedly adding duplicate tasks.
Sources and release context
This guide describes the Flux 2.6.6 documentation. Check the current release record for package conditions.
