LlamaIndex Index Processing Failure Detection

critical

reliabilityUpdated Feb 23, 2026

LlamaIndex indexes may silently fail document processing with files stuck in ERROR or PENDING states, causing incomplete search results without alerting when no successful files exist.

Sources

Index Status Monitoring | LlamaIndex Python Documentationdevelopers.llamaindex.ai

Technologies:

LlamaIndexSymptoms of this issue are visible in LlamaIndex metrics and logs

llamaindex.index.files.success_count

llamaindex.index.files.error_count

llamaindex.index.files.pending_count

llamaindex.index.files.total_count

How to detect:

Track file status counts via the LlamaCloud API: alert when SUCCESS count = 0 and ERROR count > 0, or when PENDING count remains non-zero beyond expected processing time. Monitor total_count vs. sum(SUCCESS + ERROR + PENDING) for data consistency.

Recommended action:

Poll the /pipelines/{pipeline_id}/files/status-counts endpoint every 5-10 seconds during ingestion. Alert when no files succeed within expected time window. Implement progressive monitoring: start queries only when success_count > 0, log errors with document format/size analysis when error_count spikes.