Retry-After Is a Waiting Instruction: Test Both HTTP Delay Formats
An industrial data uploader receives a response asking it to wait before trying again. Treating every Retry-After value as the same kind of timestamp can produce immediate retries or an unexpectedly long pause. Parse the documented format, retain queued records and verify when the next permitted attempt should occur.

Original conceptual illustration; not a specific product, live software screen or customer installation.
Agree the actual external interface and retry behaviour before integration testing. Start with the ProSight Software platform overview. Confirm the exact equipment, interface and project scope before selection.
Key takeaways
Retry-After can contain delay seconds or an HTTP date.
Waiting policy and safe repetition of a write are separate decisions.
Malformed or impractical instructions need a documented fallback.
1. Identify the response and its contract
RFC 9110 defines Retry-After as either an HTTP date or a non-negative integer delay in seconds after receipt of the response. Its meaning also depends on the response context. Record the status and relevant headers together rather than inspecting one isolated string.
RFC 6585 permits a 429 Too Many Requests response to include Retry-After. A service may also use the field with 503 Service Unavailable. Confirm the actual endpoint documentation; do not assume that every transient response includes the header or follows an undocumented retry policy.
This guide describes an uploader acceptance question. It does not claim that a specific retry control or endpoint is available in ProSight Software. Confirm the deployed integration method and supported service behaviour with the project team.
2. Keep relative delay and calendar time separate
For a delay-seconds value, begin with the documented response-receipt event and calculate the wait using an appropriate elapsed-time source. A bare value of 120 describes a two-minute wait, not an epoch timestamp and not 120 milliseconds.
For an HTTP date, use a proper HTTP-date parser and a trustworthy time reference. A workstation clock that is materially wrong can misinterpret an absolute target time. Record the clock assumption and permitted uncertainty rather than silently treating a past-looking date as permission for a burst.
After establishing a permitted next-attempt time, ensure other local policies do not cause an earlier retry. Where the server’s requested wait exceeds a project’s operating budget, pause and escalate under the agreed policy instead of shortening the wait without explanation.
3. Work through two hypothetical responses
In one synthetic test, a response arrives at 10:00:00 UTC with Retry-After set to 120. The next attempt must not occur before 10:02:00 under the stated receipt-time boundary. Use a suitable elapsed-time check to demonstrate the two-minute wait.
In another test, the field contains a valid HTTP date corresponding to 10:05:00 UTC. Establish the remaining interval using the trusted time basis. The two values are different encodings of waiting information; their parsing paths should both be exercised.
Test missing, malformed, zero and impractically large values using the endpoint’s supported policy. Do not improvise an unbounded immediate retry loop when parsing fails. These are illustrative acceptance cases rather than a promised service response.

Original explanatory diagram. Examples are hypothetical; verify the actual equipment, interface contract and project requirements.
4. Protect the queue during the wait
Retain record identity, original observation time and quality while uploads are delayed. A later transmission time must not silently replace the measurement time. Set an explicit queue capacity, retention rule and observable overflow outcome for the selected implementation.
The waiting instruction does not resolve whether an earlier write was stored. Handle unknown write outcomes through the endpoint’s documented acknowledgement and deduplication method. Repeating a write safely is a separate semantic question from deciding how long to wait.
Coordinate retries across workers and gateways where they share an account or endpoint limit. A single compliant worker can still participate in excessive aggregate traffic if every other worker starts independently. Validate any backoff and load-spreading policy against the server instruction.
Capture status, headers and the response-receipt boundary.
Test delay seconds and an absolute HTTP date separately.
Validate missing or invalid instructions and queue limits.
Verify recovery timing, record identity and observation timestamps.
5. Verify timing and recovery together
Use a controlled test endpoint or supported simulation so that commissioning does not overload a real service. Capture response receipt, parsed instruction, scheduled attempt, actual attempt and queue depth. Check that network recovery does not flush the entire backlog without the agreed limits.
Inspect stored records after recovery for omissions, duplicates and preserved observation times. A quieter error log is insufficient if queued data was discarded. Keep the failure and recovery cases in the handover evidence.
Bring ProSight Software the endpoint contract, a sanitised response and the uploader’s queue and retry requirements. Confirm what is available in the actual deployment before selecting controls. The next useful step is a repeatable waiting-and-recovery test with defined limits and a clear owner for unresolved cases.
Related guide and project resources
Application selection and next steps
Recommended equipment / related solutions
For this project, review ProSight Software platform overview. Use the article’s checks as acceptance questions; the listing alone does not establish the behaviour of every connected device or integration.
For an alternative project scope, review ProSight Custom project scope. Ask ProSense to confirm the required inputs, data handling and deployment options. This is a selection starting point, not a compatibility guarantee.
Need help with your application?
Send the relevant equipment documentation, field definitions and a sanitised example of the behaviour you need to verify. Confirm the available functions and integration requirements for the exact equipment and deployment.

.png)
Comments