v0.7.1

Summary

This version introduces several reliability and compatibility improvements to the proxy, focused on HTTPS support, config robustness, and downstream interoperability.

Key Changes

• HTTPS support without OpenSSL
  Migrated the HTTP client to use a Rust-native TLS backend (hyper-rustls or equivalent), removing the dependency on OpenSSL. This greatly improves build portability, especially for Alpine and musl-based containers, and eliminates the need for system OpenSSL libraries.

• Robust config: path_rules is now optional
  Fixed a bug where omitting the path_rules section in config.yaml caused a panic or crash at startup. The config loader now defaults to an empty list if not specified, using only the default latency threshold when no rules are present.

• Header filtering for downstream compatibility
  Requests forwarded by the proxy now exclude certain headers such as accept-encoding. This avoids issues with downstream servers that may not support compressed response streams (such as "Empty reply from server" errors), ensuring better compatibility with a broader range of HTTP backends.

---

Additional Improvements

• Enhanced tracing and error reporting for debugging and operations.

v0.6.0

🚀 CacheBolt v0.6.0 - Admin Port Separation & Config Enhancements

This release introduces multi-port server separation for better isolation between the proxy logic and the admin/metrics interface, along with configuration improvements that enhance maintainability and deployment flexibility.

---

✨ Features

🧩 Admin UI & Metrics on a Separate Port

CacheBolt now exposes the admin panel (/admin) and Prometheus metrics (/metrics) on a dedicated HTTP server and port, separate from the main proxy.

✅ Benefits:

• Isolate internal endpoints from public traffic
• Allow firewalled/internal-only access to admin tools
• Enable separate monitoring via sidecars or scrape configs

🔧 Configuration (in config.yaml):

proxy_port: 3000     # Main proxy port
admin_port: 3001     # Admin UI and /metrics

v0.5.0

🚀 CacheBolt v0.5.0 — Client-Controlled Cache Bypass

✨ New Features

• Header-based Cache Bypass
  Clients can now force bypass of all cache layers by including either of the following headers:
  • Cache-Control: no-cache
  • X-Bypass-Cache: true
    When present:
  • The cache is skipped (no memory or backend read).
  • The response is still stored unless explicitly disabled.
  • Useful for debugging, cache revalidation, or time-sensitive requests.

📈 Observability

• Added log:
  ⏩ Cache bypass activated for '/path' due to client header
• New Prometheus metric:
  cachebolt_bypass_cache_total

🛠️ Internal Improvements

• Normalized ignored_headers now includes bypass-related headers to avoid polluting cache keys:
  • cache-control
  • x-bypass-cache
  • x-refresh-cache

v0.4.2

Fix to the version v0.4.0

v0.4.0

This release introduces a lightweight and elegant Web UI to inspect the in-memory cache of CacheBolt. It's built as a static site and intended as a minimal dashboard for visibility and inspection.

✨ Features

🌐 Web Admin Panel (/cb-admin)

• Built using Astro
• Zero backend dependencies – purely static frontend

🗂️ Structure

• Accessible at: /cb-admin
• Custom styling with no UI frameworks for maximum performance

v0.3.1

Added

• GET /admin/status-memory: new endpoint to inspect in-memory cache entries. Returns key, path, inserted_at, size_bytes, and expires_in_secs.
• TTL-based expiry metadata is now computed and exposed from memory cache entries using cache.ttl_secs from config.

Changed

• Moved cache invalidation endpoint from DELETE /cache to DELETE /admin/cache for consistency with admin operations.
• Cleaned up redundant timeout logic in proxy layer. Latency-based failover now exclusively relies on latency_failover rules from the config.

Fixed

• Corrected config field inconsistencies to ensure TTL and memory thresholds are parsed and enforced properly.

v0.3.0

✨ Features

• Probabilistic Cache Refreshing (cache.refresh_percentage)
  • New configuration option under the cache section:

    cache:
      refresh_percentage: 10

  • Allows a percentage of requests to bypass the cache and fetch a fresh response from the backend.
  • Example: 10 means ~1 in every 10 requests to the same key will force a refresh.
  • Helps prevent stale data and reduces the need for manual invalidation.
  • Automatically re-stores the refreshed result in memory and persistent storage.

🐛 Fixes

• Fixed handling of root path /
  • The proxy now correctly forwards requests to / (previously may have failed or misrouted).
  • Fully compatible with downstream services expecting root-level access.

📊 Metrics & Internals

• Added debug logs and counters for refresh-triggered requests.
• cachebolt_memory_hits_total no longer counts requests that triggered a refresh.
• Refresh logic respects concurrency limits and latency-based failover rules.

🔄 Behavior

• If refresh_percentage is set to 0 or omitted, the cache will behave deterministically (no automatic refresh).
• Refresh logic is key-specific and probabilistic based on request counts.

v0.2.0

📦 CacheBolt v0.2.0 (2025-06-11)

✨ Added

• Global cache invalidation via DELETE /cache?backend=true
  • Clears all in-memory entries
  • Deletes all objects from the configured persistent backend (S3, GCS, Azure, Local)
• Support for structured JSON error responses on invalid query usage
• Prometheus metrics now reflect full deletion operations when backends are flushed

🔧 Changed

• Removed pattern-based deletion from the /cache endpoint
  • Simplified the behavior: you now either clear everything, or nothing
  • Backend deletion only triggers if ?backend=true is explicitly set

📝 Documentation

• Updated README.md:
  • Added new section: 🧹 Cache Invalidation
  • Described usage of DELETE /cache?backend=true and example scenarios

🧪 Internal

• Improved logging around deletion success/failure per storage provider
• Hardened concurrency and error handling in backend deletion routines

---

🚀 CacheBolt continues to evolve as a high-performance reverse proxy with multi-cloud caching support.

v0.1.0

Merge pull request #3 from msalinas92/feature/prometheus-metrics

feat: prometheus-metrics