We are trying to use the Pipeline Monitoring API in our implementation to trace which of the service calls have failed,
using e.g. ___https://elastic.snaplogic.com/api/1/rest/public/runtime/Partners?last_hours=100&state=Completed,Failed,Stopped.
Based on the output/results of the Monitoring API, we’ll build a exception replay service to pick the document data (message payload) of all the failed transactions and replay/re-trigger then from the pipeline.
But the challenge is when we use error view in the pipeline, the state of a particular transaction is always captured as ‘Completed’ instead of ‘Failed’ or ‘NoUpdate’, 'Prepared’, 'Started’, 'Queued’, 'Stopped’, 'Stopping’. Also, so far we’re not able to figure out how to problematically overwrite the state as well, which if possible will come handy to filter the ‘failed’ transactions for replay purpose.
Please suggest if someone has observed similar design challenge or there is a possible workaround available.