v0.5.0

What's Changed

• Convert to python by @hev in #10
• Schema by @hev in #11
• Remove .agent and .beads from tracking by @hev in #12

Full Changelog: v0.4.0...v0.5.0

v0.4.0

What's Changed

Bug Fixes

• Fixed exporter query costs - Recall estimation is now opt-in to prevent unexpected query charges (#9, fixes #8)

New Features

• Added --include-recall flag for exporter (default: disabled)
• Added --recall-interval option (default: 3600s / 1 hour) for separate recall refresh scheduling
• Recall data is now cached between refresh intervals

Cost Impact

Previously, running the exporter with default settings could generate ~72,000 query calls per weekend (with 10 namespaces). Now:

• Default (recall disabled): 0 query calls
• Opt-in with hourly refresh: ~1,200 query calls

Docker

Updated hevmind/tpuff-exporter container available.

v0.3.0

Installation

npm install -g tpuff-cli

Docker Images

# Embeddings service
docker pull hevmind/tpuff-embeddings:0.3.0

# Prometheus exporter
docker pull hevmind/tpuff-exporter:0.3.0

What's Changed

• readme cleanup by @hev in #4
• Add recall estimation feature to CLI, exporter, and dashboard by @hev in #5

Full Changelog: v0.2.0...v0.3.0

v0.2.0

New Features

Prometheus Exporter & Monitoring Stack

• Prometheus exporter for namespace metrics (tpuff export command)
  • Exports namespace rows, storage, unindexed bytes, and index status
  • Supports single region or all-regions monitoring
  • Configurable refresh intervals and timeouts
  • Available as Docker image: hevmind/tpuff-exporter:0.2.0

Grafana Dashboard

• Pre-built Grafana dashboard for visualizing Turbopuffer metrics
  • 2GB unindexed bytes threshold alerts with color-coded warnings
  • Aggregate metrics with drill-down filtering by region/namespace
  • Per-namespace breakdown with sortable table
  • Regional distribution charts

Prometheus Alert Rules

• Pre-configured alert rules for proactive monitoring
  • Warning alert: >2GB unindexed data (fires after 5 minutes)
  • Critical alert: >5GB unindexed data (fires after 10 minutes)
  • Infrastructure health alerts for low index health
  • Exporter health monitoring
  • Ready for Slack, PagerDuty, or email integration

Enhanced Delete Command

• Added --prefix flag to delete all namespaces matching a prefix
• Improved confirmation prompts with namespace counts
• Debug logging for troubleshooting deletion operations

Improvements

• Enhanced list command with better metadata fetching and error handling
• Added debug logging throughout for easier troubleshooting
• Comprehensive documentation for Docker deployment and monitoring
• Complete observability stack with docker-compose examples

Docker Images

• hevmind/tpuff-embeddings:0.2.0 - Python embedding service
• hevmind/tpuff-exporter:0.2.0 - Prometheus exporter (new)

Installation

npm install -g tpuff-cli

Documentation

• See https://github.com/hev/tpuff/blob/main/DOCKER.md for complete monitoring stack setup
• See https://github.com/hev/tpuff/blob/main/grafana/README.md for dashboard usage
• See https://github.com/hev/tpuff/blob/main/prometheus/README.md for alert configuration

v0.1.0

Installation

npm install -g tpuff-cli

Docker Image

docker pull hevmind/tpuff-embeddings:0.1.0

Full Changelog: https://github.com/hev/tpuff/commits/v0.1.0