Open code, practice, and notes →
Download source with English comments · for reading
Web Workers · 20 Failure Recovery
Series contents · Technology map
Stage: Queues and Error Recovery · Editable example · Storage ID: web-workers-step-20
Goals and learning outcomes
- Learning outcome: explain the role of Worker failure. Input cases: a normal task and a failing task.
- Deliverable: replacement count, final error, and completed IDs. Explain the rationale for handling Retry, linking it to the code.
Prerequisites
Prepare the experiment results for Time limit and Single settlement from Unit 19, “Timeouts and Duplicate Responses”.
Key concepts and explanations
Worker failure
An error event may affect multiple in-progress tasks. Find the requests assigned to that Worker.
Retry
Retry side-effect-free tasks with care. Without an attempt limit, failures can form a loop.
Practice setup
First check the execution scope guidance on the example screen. The following describes the environment needed to extend the assignment in a separate project. The existing web-workers-01·02·03 are starting examples. Implement the new assignments in main.js and worker.js in a separate local HTTP project. Preserve CSP and iframe isolation. For SharedArrayBuffer, check COOP and COEP requirements on an external server you control; use the transfer path if they are not met. Measure performance yourself.
The linked example is a starting point. It does not implement every task below. Complete work beyond its supported scope in a separate project.
Practice tasks
- In src/day-20.js in a separate practice project, trigger an error for a specific input and implement pool replacement and one retry. Refer to the existing Unit 1·14·28 examples only as starting points.
- Input conditions: a normal task and a failing task. Baseline: 0 retries. Run under the baseline conditions, then record observations and errors in notes/day-20.md.
- Variable: Retry limit; changed value: 1 retry. Run again through the same input path. Deliverable to save: replacement count, final error, and completed IDs. Hypothesis to check: Processing ends with a result or error without endless recreation. Compare it with the actual results, and record whether they agree and why they differ.
Variables to observe
| Variable | Baseline | Change | Expected change / hypothesis |
|---|---|---|---|
| Retry limit | 0 retries | 1 retry | Processing ends with a result or error without endless recreation. |
Predictions in this table are not measured results. Change one condition at a time and record why the actual result differs.
Completion criteria
- Required deliverable: replacement count, final error, and completed IDs. Results are available for both the baseline and changed conditions, allowing comparison of 2 conditions; input values and the execution environment are recorded together.
- The explanation of differences between the two conditions connects both concepts: Worker failure and Retry. For unexpected items, at least 1 cause or reason they remain unresolved is recorded.
Self-check
- Which processing stage does Worker failure affect in this exercise? Input cases: a normal task and a failing task.
- Variable: Retry limit. Hypothesis to check: Processing ends with a result or error without endless recreation. Did the actual results agree with the hypothesis? Explain the evidence in terms of Retry.
Official references
Linked example and execution scope
Example source · Runtime environments and limits
On a request-error response, the supplied example replaces the Worker and submits a new [9,9] sum request with a new ID. It does not implement Worker error-event recovery or retry the original request.
Source code and records are not overwritten automatically. Resetting an existing example also preserves your notes.
Next learning
Previous: Timeouts and Duplicate Responses · Next: Separating MessageChannels