Understanding RAG Pipeline Status
Your RAG pipeline goes through different states as it processes your data. This page explains what each status means and what to expect during normal operation.
CREATED
Initial status when a pipeline is first created. Your pipeline is being set up.
DEPLOYING
The pipeline has been created, and is now being deployed.
DEPLOYED
Setup is complete and your pipeline is ready to begin processing data.
ERROR_DEPLOYING
An error occurred during deployment.
BACKFILLING
Your pipeline is performing its initial load of data into the vector database. This continues until no new data has been added for 30 seconds.
LISTENING
Your pipeline is active and monitoring your data sources for changes, but no data is currently being processed.
PROCESSING
Your pipeline has processed new data within the last 30 seconds. During normal operation, your pipeline will switch between LISTENING and PROCESSING as changes are detected.
SHUTDOWN
Your pipeline has been manually stopped. The retrieval endpoint still works, but your pipeline won't process any new changes from your data sources.
STARTED
Your pipeline is starting up after being shut down. It will quickly move to either LISTENING or PROCESSING.
OVERQUOTA
Your pipeline has exceeded its data processing limit for the current month.
HIBERNATING
Your pipeline has been inactive for 14 days (no data, no activity, no use of the retrieval endpoint). It needs to be manually restarted. If it remains inactive after restart, it will hibernate again.
Last updated