When a measurement submission fails, we send it straight into S3. This usually happens during incidents, and it's a good worst-case scenario but it requires reprocessing the missing data.
A better best-effort thing for this kind of situation is to fall back to a second fastpath
In this context, the requires tasks are the following:
When a measurement submission fails, we send it straight into S3. This usually happens during incidents, and it's a good worst-case scenario but it requires reprocessing the missing data.
A better best-effort thing for this kind of situation is to fall back to a second fastpath
In this context, the requires tasks are the following:
collector_idin this fastpath is different to the original oneooniprobeservice to fallback to the second fastpath if it failsOTHER_COLLECTORSenv variable inoonimeasurementsto also include the second fastpath. This is important to support real-time serving of measurements from the spool dir