Row Store Metrics

This page documents Prometheus metrics related to the row store system in Yellowbrick.

Purpose

These metrics help track the internal state, storage usage, and operational status of the row store layers, including:

• File counts for different Yellowbrick row store components (data, commit, delete, unused)
• Accumulated unflushed data and total bytes stored
• System health and status transitions
• Row store table counts and utilization thresholds

They are useful for debugging catalog performance issues, tracking file churn, and monitoring for cleanup/backlog risks in the transactional metadata layer.

Metrics

Name	Type	Freq	Labels	Description
yb_catalog_rowstore_bytes_used	gauge	15m	ro_threshold_bytes, block_threshold_bytes	Total number of bytes used in database catalog row store
yb_rowstore_status	gauge	15m	status	Row store status as labels: NORMAL, WARNING, READ_ONLY, YRS_FULL, FULL, or ABNORMAL; 1 if active, else 0.
yb_yrs_rowstore_active_flushes	gauge	15m	-	Number of Yellowbrick row store tables with unflushed bytes
yb_yrs_rowstore_commit_file_count	gauge	15m	-	Total number of Yellowbrick row store commit files
yb_yrs_rowstore_data_file_count	gauge	15m	-	Total number of Yellowbrick row store data files
yb_yrs_rowstore_delete_file_count	gauge	15m	-	Total number of Yellowbrick row store delete files
yb_yrs_rowstore_table_count	gauge	15m	-	Number of row Yellowbrick store tables
yb_yrs_rowstore_unused_file_count	gauge	15m	-	Total number of Yellowbrick row store unused files