Classify the symptom before changing settings
| Symptom | First comparison |
|---|---|
| No quote | Source session and instrument entitlement versus the receiving connection. |
| Old quote | Source timestamp and reception sequence at each observable stage. |
| Unexpected price or spread | Upstream value versus transformations and the intended instrument definition. |
| Only some clients affected | Visible account or group configuration versus an unaffected test case. |
Use an unaffected instrument or client only when its configuration is genuinely comparable. Different sessions or contracts can explain an apparent inconsistency.
Trace reception, transformation and delivery
Check connection state and the availability of the intended source instrument. Then review symbol aliases, decimals, session assumptions and any transformations made by the integration. Compare the observed configuration with the approved mapping record, not an operator’s memory.
Collect a small time-bounded sample with source and reception timestamps where available. Clock differences must be accounted for before a delay is assigned to one component. Keep original observations available while changes are tested.
Restore a known condition and verify it
Agree a reversible configuration change or a provider investigation based on the failing boundary. Test the affected instrument and a suitable control afterwards, including session transitions when relevant. Record the result and update the mapping or monitoring record.
If a fallback source is proposed, validate its instrument and transformation assumptions separately. Availability of another source does not prove price equivalence or contractual permission to use it. Bring the evidence to a price-feed integration review instead of substituting an untested source in production.