> ## Documentation Index
> Fetch the complete documentation index at: https://www.octoparse.com/docs/llms.txt
> Use this file to discover all available pages before exploring further.
# Checkpoint resume
> Resume a supported Octoparse local run from saved progress instead of recollecting completed work from the beginning.
Checkpoint resume lets an interrupted local extraction continue from saved progress instead of recollecting the entire task from the beginning.
Introduced in Octoparse 10.2, it is designed for long-running local tasks where restarting could mean revisiting many pages, repeating completed loops, and waiting for previously collected data again.
## What it saves
While a supported local task runs, Octoparse records enough progress to reconstruct where the extraction stopped. Depending on the workflow, the saved checkpoint can include:
* The number of rows already output
* The current position in a URL, text, item, or scrolling loop
* The remaining items in the loop
* The workflow position and page needed to reach the saved point
* Progress for individual subtasks in a Boost mode run
The extracted data already stored in the current run is preserved. When the task resumes, Octoparse restores that run and continues working toward the remaining data.
## What happens when you resume
Suppose a local task is collecting 5,000 product pages and stops after producing 3,200 rows.
When you open or start the task again, Octoparse checks whether the saved checkpoint still belongs to the current task and can be used. If it is valid, the runner shows the recorded progress and gives you two choices:
* **Resume** restores the previous run, returns to the saved workflow position, and continues with the remaining work.
* **Re-collect** clears the checkpoint and starts a new run from the beginning.
Resuming does not always mean opening the last URL and immediately extracting the next row. Some workflows must replay part of their navigation or loop state to safely reach the saved position. During this replay, Octoparse uses the recorded progress to skip completed work and avoid treating replayed output as new data. Once the saved position is matched, normal extraction continues.
In Boost mode, Octoparse also restores progress by subtask: completed subtasks are skipped, while paused, stopped, failed, or unfinished subtasks return to the pending queue.
## How much work it can preserve
The benefit depends on when the run stopped and how much work had already completed.
| Without Checkpoint resume | With Checkpoint resume |
| --------------------------------------------------- | -------------------------------------------------- |
| Restart the task from its first input or page | Continue toward the unfinished portion of the run |
| Revisit pages and loops that were already processed | Restore recorded loop and workflow progress |
| Wait for completed subtasks to run again | Skip completed Boost mode subtasks |
| Create a new run after every interruption | Continue using the existing run and collected data |
The improvement is greatest for large URL lists, deep pagination, long loops, or Boost mode tasks with many subtasks. A run interrupted near the beginning will preserve less work than one interrupted near completion.
## Resume an interrupted local run
Open the interrupted local run or start the same task from the task list.
If a valid checkpoint is available, Octoparse displays the saved position. This may include output rows, loop progress, or completed and remaining subtasks.
Select **Resume** to restore the previous run. Choose **Re-collect** when you intentionally want to discard the saved progress and run the task from the beginning.
Octoparse may replay page navigation or workflow actions before it reaches the recorded position. The runner indicates when it is restoring progress and when the checkpoint has been matched.
After the remaining extraction finishes, check the output row count and sample records around the interruption point. Confirm that the final dataset covers the intended inputs without unexpected missing or repeated records.
## When a checkpoint cannot be used
A checkpoint represents a specific task configuration and run. Octoparse validates it before resuming so that saved progress is not applied to an incompatible task.
You may need to re-collect when:
* The task or its input URLs were changed after the checkpoint was created
* The checkpoint is missing, damaged, or cannot be matched to the current workflow
* The saved progress came from a different local execution mode
* The previous run completed successfully and has no remaining work
* A scheduled local run starts; scheduled runs begin fresh rather than waiting for a resume decision
If validation fails, Octoparse marks the checkpoint as unavailable and offers a fresh run instead of silently continuing from an uncertain position.
Checkpoints are saved with the local desktop run. Resume the task in the same Octoparse desktop environment where the progress was recorded.
## Checkpoint resume, pause, and retry
These actions solve different problems:
| Action | Use it when |
| ----------------- | ----------------------------------------------------------------------------------------------------------- |
| Pause and resume | The runner is still open and you want to temporarily pause the active process |
| Checkpoint resume | The previous local run stopped and must be reconstructed from saved progress |
| Re-collect | The task changed, the previous progress is no longer useful, or you want a clean dataset from the beginning |
## Related pages
Choose where a task should run.
Understand sequential and parallel local extraction.
Review task progress and execution results.