Changelog
version 0.1.2 (2025.07.14)
Store Configuration Enhancement
New Environment Variables
EX_ESDB_STORE_DESCRIPTION
: Human-readable description of the store for documentation and operational purposesEX_ESDB_STORE_TAGS
: Comma-separated tags for store categorization and filtering (e.g., "production,cluster,core")
Rich Store Configuration
- Operational Metadata: Added
created_at
,version
, andstatus
fields to store configuration - Resource Management: Added
priority
andauto_start
fields for resource allocation control - Administrative Info: Enhanced store registry with
tags
,environment
, anddescription
fields - Enhanced Querying: Store registry now supports filtering by tags, environment, priority, and other metadata
Configuration Integration
- Runtime Configuration: New environment variables integrated into
config/runtime.exs
- Options System: Added parsers in
ExESDB.Options
with comma-separated tag parsing - Store Registry: Enhanced
build_store_config/1
to include all new metadata fields - Development Environment: Updated
dev-env/*.yaml
files with new environment variables
Architectural Refactoring
NotificationSystem Introduction
- New Core Component: Created
ExESDB.NotificationSystem
as a core supervisor for event notification - Leadership Integration: Moved
LeaderSystem
from clustering layer to core system - Core System Enhancement: Updated
CoreSystem
to includeNotificationSystem
alongsidePersistenceSystem
andStoreSystem
- Supervision Order:
PersistenceSystem
→NotificationSystem
→StoreSystem
for proper dependency management
LeaderWorker Availability Fix
- Core System Integration: LeaderWorker now starts as part of core system, not clustering components
- Startup Order: LeaderWorker is available before clustering components attempt to use it
- Resolved
:noproc
Error: Fixed LeaderWorker activation failures by ensuring it's always running when needed - Single and Cluster Mode: LeaderWorker now available in both single-node and cluster modes
System Architecture Cleanup
- Removed LeadershipSystem: Consolidated functionality into
NotificationSystem
- Cleaner Separation: Core functionality (leadership, events) vs clustering (coordination, membership)
- Improved Documentation: Updated supervision tree documentation to reflect new architecture
- Simplified Dependencies: Reduced coupling between core and clustering components
Process Management Enhancement
Graceful Shutdown Implementation
- Universal Coverage: All 18 GenServer processes now implement graceful shutdown
- Terminate Callbacks: Added
terminate/2
callbacks to all GenServers for proper cleanup - Exit Trapping: Enabled
Process.flag(:trap_exit, true)
on all GenServers - Resource Cleanup: Proper cleanup of Swarm registrations, PubSub subscriptions, and Khepri stores
Enhanced Process Lifecycle
- Swarm Registration Cleanup: Worker processes properly unregister from Swarm on shutdown
- PubSub Subscription Cleanup: EventProjector properly unsubscribes from topics
- Khepri Store Shutdown: Store processes gracefully stop Khepri instances
- Network Monitoring: NodeMonitor properly disables network monitoring on shutdown
Development Environment
Configuration Updates
- proc-sup Configuration: Added description "Process Supervisor Event Store" and tags "development,cluster,proc-sup,core"
- reg-gh Configuration: Added description "Registration System Event Store" and tags "development,cluster,reg-gh,registration"
- Environment-Specific Tags: Different tags for development vs production environments
- Consistent Formatting: Standardized environment variable layout across all configuration files
Benefits
Operational Improvements
- Enhanced Monitoring: Rich store metadata enables better operational visibility
- Improved Debugging: Store descriptions and tags help identify issues faster
- Better Resource Management: Priority and auto-start fields enable fine-grained control
- Cleaner Shutdown: All processes terminate gracefully without resource leaks
Development Experience
- Clearer Architecture: Separation of core vs clustering concerns
- Consistent Configuration: Standardized environment variable management
- Better Testability: Core components can be tested independently of clustering
- Simplified Debugging: LeaderWorker availability issues resolved
System Reliability
- Reduced Race Conditions: Proper startup order prevents timing-related failures
- Resource Leak Prevention: Graceful shutdown prevents resource accumulation
- Improved Fault Tolerance: Better separation of concerns reduces cascade failures
- Enhanced Observability: Rich metadata supports better monitoring and alerting
version 0.1.1 (2025.07.13)
StoreRegistry Refactoring
Enhanced Architecture
- Store Registration Centralization: Moved store registration functionality from
StoreCluster
to dedicatedStoreRegistry
module - Self-Registration:
StoreRegistry
now automatically registers its own store during initialization whenstore_id
is provided - Simplified StoreCluster:
StoreCluster
now focuses purely on cluster coordination without store registration concerns
API Integration
- ExESDBGater.API Integration:
list_stores()
function now directly callsExESDB.StoreRegistry.list_stores()
instead of maintaining local state - Single Source of Truth: Store information is now centralized in
StoreRegistry
across the entire system - Improved Error Handling: Added proper error handling for StoreRegistry calls in GaterAPI
System Integration
- StoreSystem Supervision: Added
StoreRegistry
to theStoreSystem
supervisor with proper startup order - Component Isolation: Each component now has a single, well-defined responsibility
- Cleaner State Management: Removed redundant store state from multiple components
Benefits
- Separation of Concerns: Clear boundaries between clustering and registration responsibilities
- Maintainability: Easier to maintain and reason about store registration logic
- Testability: Store registration can now be tested in isolation
- Reduced Coupling: Components are less tightly coupled and more modular
version 0.0.17 (2025.07.01)
Auto-Clustering
ExESDB
nodes now automatically join the cluster- "Split-Brain" scenarios are now mitigated
BCUtils
- All functionality related to styling is now transferred to the
:bc_utils
package. - Added a Banner after startup.
- Logger filtering for Swarm and LibCluster noise reduction (via BCUtils.LoggerFilters)
ExESDB Logger Filtering
Features
The ExESDB.LoggerFilters
module provides additional log noise reduction specifically for ExESDB's distributed systems components:
- Ra Consensus Filtering: Reduces Ra heartbeat, append_entries, pre_vote, request_vote, and routine state transition messages while preserving all errors/warnings
- Khepri Database Filtering: Filters internal Khepri operations (cluster state, store operations) at info/debug levels while maintaining error/warning visibility
- Enhanced Swarm Filtering: Complements BCUtils filtering with additional ExESDB-specific Swarm noise reduction
- Enhanced LibCluster Filtering: Complements BCUtils filtering with additional ExESDB-specific cluster formation noise reduction
Benefits
- Dramatically improves log readability in development and production environments
- Intelligent filtering preserves all error and warning messages
- Focused on ExESDB-specific distributed systems infrastructure (Ra, Khepri)
- Works in conjunction with BCUtils.LoggerFilters for comprehensive noise reduction
ExESDBGater
- The
ExESDB.GatewayAPI
is moved to the:ex_esdb_gater
package.
Features
Snapshots Subsystem provides cluster wide support for reading and writing snapshots, using a key derived from the source_uuid
, stream_uuid
and version
of the snapshot.
version 0.0.16 (2025.06.26)
Snapshots
Features
Snapshots Subsystem provides cluster wide support for reading and writing snapshots, using a key derived from the source_uuid
, stream_uuid
and version
of the snapshot.
record_snapshot/5
functiondelete_snapshot/4
functionread_snapshot/4
functionlist_snapshots/3
function
Supported by Gateway API
It is advised to use ExESDB.GatewayAPI
to access the Snapshots Subsystem.
version 0.0.15 (2025.06.15)
Subscriptions
Transient subscriptions
:by_stream
,:by_event_type
,:by_event_pattern
,:by_event_payload
- Events are forwarded to
Phoenix.PubSub
for now
Persistent subscriptions
:by_stream
, with support for replaying from a given version- Events are forwarded to a specific subscriber process
ack_event/3
function is provided
"Follow-the-Leader"
Emitter processes are automatically started on the leader node, when a new leader is elected.
Gateway API
- A cluster-wide gateway API is provided
- is an entry point for all the other modules
- provides basic High-Availability and Load-Balancing
version 0.0.9-alpha (2025.05.04)
Subscriptions
ExESDB.Subscriptions
modulefunc_registrations.exs
file- emitter trigger in
khepri
now only uses theerlang
-native :pg library (process groups)
Skeleton support for Commanded
ExESDB.Commanded.Adapter
moduleExESDB.Commanded.Mapper
module
version 0.0.8-alpha
2025.04.13
- Added
ExESDB.EventStore.stream_forward/4
function - Added
BeamCampus.ColorFuncs
module - Added
ExESDB.Commanded.Adapter
module - Refactored
ExESDB.EventStreamReader
andExESDB.EventStreamWriter
modules: - Streams are now read and written using the
ExESDB.Streams
module - Removed
ExESDB.EventStreamReader
module - Removed
ExESDB.EventStreamWriter
module
version 0.0.7-alpha
version 0.0.1-alpha
2025.03.25
- Initial release