faye-redis-ng 1.0.9 → 1.0.10

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: cd86d6fdb530405ff0dd146705d45d9c665cfa9f778987ead89d4bb497db06eb
-  data.tar.gz: ddc3a783e0007e452d69eb97e6f2540a089427420119a4414e91281e1b4a9d9d
+  metadata.gz: 055d9be802f284752c5ae672d4d11c9121ef023739e17508e9957ef5d6880df7
+  data.tar.gz: 1119068a05ad0d0dbfffdd7d92cb1b882d9f98308f2a868818e341e01ed21139
 SHA512:
-  metadata.gz: c2a3f9d350a9aeab02bcfc6589a6f3e81e2973396d54397510d46787e1d7cde81de064a97f44428bf3534f266ccda104ae1535fd8108197e0862aeae1c28b71e
-  data.tar.gz: 4f41d51dae3a17b7d44171384f6a2646ac7492ae68c4f85845feac100f20d59f609105ebfc90d5fdffd415e893cb3e7556bb4ade191da84d867e16a90e8bf664
+  metadata.gz: 1e55a67832698a6969390882eaf687c7829de4f70907bc33e796103f9336cc403879bf6721f052611c3073f64650822ef7d12e8acc13a3ec1f6b55b9f11b1810
+  data.tar.gz: 9a46f1bd0136923f320d723cf96cf7e64cd7193f9f4c2e19704bed4b268e180892b1ae54427a5888516816597d9ee77170a82cb6784203eb22ba5c11cd456639

data/CHANGELOG.md

@@ -7,6 +7,51 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+## [1.0.10] - 2025-10-30
+
+### Fixed - Critical Memory Leak (P0 - Critical Priority)
+- **TTL Reset on Every Operation**: Fixed message queue and subscription TTL being reset on every operation
+  - **Problem**: `EXPIRE` was called on every `enqueue` and `subscribe`, resetting TTL to full duration
+  - **Impact**: Hot/active queues and subscriptions never expired, causing unbounded memory growth
+  - **Solution**: Implemented Lua scripts to only set TTL if key has no TTL (TTL == -1)
+  - Message queues: `enqueue` now checks TTL before setting expiration
+  - Subscriptions: `subscribe` now checks TTL before setting expiration on all keys:
+    - `faye:subscriptions:{client_id}` (SET)
+    - `faye:channels:{channel}` (SET)
+    - `faye:subscription:{client_id}:{channel}` (Hash)
+    - `faye:patterns` (SET)
+  - **Impact**: Prevents memory leak for active clients with frequent messages/re-subscriptions
+
+- **Orphaned Message Queue Cleanup**: Added dedicated cleanup for orphaned message queues
+  - Added `cleanup_orphaned_message_queues_async` to scan and remove orphaned message queues
+  - Added `cleanup_message_queues_batched` for batched deletion with EventMachine yielding
+  - Integrated into `cleanup_orphaned_data` workflow (Phase 3)
+  - **Impact**: Ensures message queues are cleaned up even if subscription cleanup misses them
+
+### Added
+- Lua script-based TTL management for atomic operations
+- Comprehensive TTL behavior tests (3 new tests):
+  - `sets TTL on first enqueue`
+  - `does not reset TTL on subsequent enqueues`
+  - `sets TTL again after queue expires and is recreated`
+
+### Changed
+- `MessageQueue#enqueue`: Uses Lua script to prevent TTL reset
+- `SubscriptionManager#subscribe`: Uses Lua script to prevent TTL reset on all subscription keys
+- `SubscriptionManager#cleanup_orphaned_data`: Added Phase 3 for message queue cleanup
+
+### Technical Details
+**Lua Script Approach**:
+```lua
+redis.call('RPUSH', KEYS[1], ARGV[1])
+local ttl = redis.call('TTL', KEYS[1])
+if ttl == -1 then  -- Only set if no TTL exists
+  redis.call('EXPIRE', KEYS[1], tonumber(ARGV[2]))
+end
+```
+
+**Test Coverage**: 213 examples, 0 failures, 87.22% line coverage
+
 ## [1.0.9] - 2025-10-30
 
 ### Fixed - Concurrency Issues (P1 - High Priority)
@@ -116,9 +161,12 @@ end
 ```
 
 ### Test Coverage
-- All 177 tests passing
-- Line Coverage: 85.77%
-- Branch Coverage: 55.04%
+- **210 tests passing** (+33 new tests, +18.6%)
+- **Line Coverage: 89.69%** (+3.92% from v1.0.8)
+- **Branch Coverage: 60.08%** (+5.04% from v1.0.8)
+- Added comprehensive tests for all P1/P2 fixes
+- Added edge case and error handling tests
+- All new features have corresponding test coverage
 
 ### Upgrade Notes
 This release includes important concurrency and performance fixes. Recommended for all users, especially:

data/lib/faye/redis/message_queue.rb

@@ -19,15 +19,19 @@ module Faye
 
         # Store message directly as JSON
         message_json = message_with_id.to_json
+        key = queue_key(client_id)
 
         @connection.with_redis do |redis|
-          # Use RPUSH with EXPIRE in a single pipeline
-          redis.pipelined do |pipeline|
-            # Add message to client's queue
-            pipeline.rpush(queue_key(client_id), message_json)
-            # Set TTL on queue (only if it doesn't already have one)
-            pipeline.expire(queue_key(client_id), message_ttl)
-          end
+          # Use Lua script to atomically RPUSH and set TTL only if key has no TTL
+          # This prevents resetting TTL on every enqueue for hot queues
+          redis.eval(<<~LUA, keys: [key], argv: [message_json, message_ttl.to_s])
+            redis.call('RPUSH', KEYS[1], ARGV[1])
+            local ttl = redis.call('TTL', KEYS[1])
+            if ttl == -1 then
+              redis.call('EXPIRE', KEYS[1], tonumber(ARGV[2]))
+            end
+            return 1
+          LUA
         end
 
         EventMachine.next_tick { callback.call(true) } if callback

data/lib/faye/redis/subscription_manager.rb

@@ -14,34 +14,48 @@ module Faye
         timestamp = Time.now.to_i
         subscription_ttl = @options[:subscription_ttl] || 86400  # 24 hours default
 
+        client_subs_key = client_subscriptions_key(client_id)
+        channel_subs_key = channel_subscribers_key(channel)
+        sub_key = subscription_key(client_id, channel)
+
         @connection.with_redis do |redis|
-          redis.multi do |multi|
-            # Add channel to client's subscriptions
-            multi.sadd?(client_subscriptions_key(client_id), channel)
-            # Set/refresh TTL for client subscriptions list
-            multi.expire(client_subscriptions_key(client_id), subscription_ttl)
-
-            # Add client to channel's subscribers
-            multi.sadd?(channel_subscribers_key(channel), client_id)
-            # Set/refresh TTL for channel subscribers list
-            multi.expire(channel_subscribers_key(channel), subscription_ttl)
-
-            # Store subscription metadata
-            multi.hset(
-              subscription_key(client_id, channel),
-              'subscribed_at', timestamp,
-              'channel', channel,
-              'client_id', client_id
-            )
-            # Set TTL for subscription metadata
-            multi.expire(subscription_key(client_id, channel), subscription_ttl)
-
-            # Handle wildcard patterns
-            if channel.include?('*')
-              multi.sadd?(patterns_key, channel)
-              # Set/refresh TTL for patterns set
-              multi.expire(patterns_key, subscription_ttl)
+          # Use Lua script to atomically add subscriptions and set TTL only if keys have no TTL
+          # This prevents resetting TTL on re-subscription
+          redis.eval(<<-LUA, keys: [client_subs_key, channel_subs_key, sub_key], argv: [channel, client_id, timestamp.to_s, subscription_ttl])
+            -- Add channel to client's subscriptions
+            redis.call('SADD', KEYS[1], ARGV[1])
+            local ttl1 = redis.call('TTL', KEYS[1])
+            if ttl1 == -1 then
+              redis.call('EXPIRE', KEYS[1], ARGV[4])
+            end
+
+            -- Add client to channel's subscribers
+            redis.call('SADD', KEYS[2], ARGV[2])
+            local ttl2 = redis.call('TTL', KEYS[2])
+            if ttl2 == -1 then
+              redis.call('EXPIRE', KEYS[2], ARGV[4])
             end
+
+            -- Store subscription metadata
+            redis.call('HSET', KEYS[3], 'subscribed_at', ARGV[3], 'channel', ARGV[1], 'client_id', ARGV[2])
+            local ttl3 = redis.call('TTL', KEYS[3])
+            if ttl3 == -1 then
+              redis.call('EXPIRE', KEYS[3], ARGV[4])
+            end
+
+            return 1
+          LUA
+
+          # Handle wildcard patterns separately
+          if channel.include?('*')
+            redis.eval(<<-LUA, keys: [patterns_key], argv: [channel, subscription_ttl])
+              redis.call('SADD', KEYS[1], ARGV[1])
+              local ttl = redis.call('TTL', KEYS[1])
+              if ttl == -1 then
+                redis.call('EXPIRE', KEYS[1], ARGV[2])
+              end
+              return 1
+            LUA
           end
         end
 
@@ -207,11 +221,14 @@ module Faye
         scan_orphaned_subscriptions(active_set, namespace) do |orphaned_subscriptions|
           # Phase 2: Clean up orphaned subscriptions in batches
           cleanup_orphaned_subscriptions_batched(orphaned_subscriptions, namespace, batch_size) do
-            # Phase 3: Clean up empty channels (yields between operations)
-            cleanup_empty_channels_async(namespace) do
-              # Phase 4: Clean up unused patterns
-              cleanup_unused_patterns_async do
-                callback.call if callback
+            # Phase 3: Clean up orphaned message queues
+            cleanup_orphaned_message_queues_async(active_set, namespace, batch_size) do
+              # Phase 4: Clean up empty channels (yields between operations)
+              cleanup_empty_channels_async(namespace) do
+                # Phase 5: Clean up unused patterns
+                cleanup_unused_patterns_async do
+                  callback.call if callback
+                end
               end
             end
           end
@@ -305,6 +322,80 @@ module Faye
         EventMachine.next_tick { callback.call }
       end
 
+      # Clean up orphaned message queues for non-existent clients
+      # Scans for message queues that belong to clients not in the active set
+      def cleanup_orphaned_message_queues_async(active_set, namespace, batch_size, &callback)
+        orphaned_queues = []
+
+        # Batch scan to avoid holding connection
+        scan_batch = lambda do |cursor_value|
+          begin
+            @connection.with_redis do |redis|
+              cursor, keys = redis.scan(cursor_value, match: "#{namespace}:messages:*", count: 100)
+
+              keys.each do |key|
+                client_id = key.split(':').last
+                orphaned_queues << key unless active_set.include?(client_id)
+              end
+
+              if cursor == "0"
+                # Scan complete, now clean up in batches
+                if orphaned_queues.any?
+                  cleanup_message_queues_batched(orphaned_queues, batch_size) do
+                    EventMachine.next_tick { callback.call }
+                  end
+                else
+                  EventMachine.next_tick { callback.call }
+                end
+              else
+                # Continue scanning
+                EventMachine.next_tick { scan_batch.call(cursor) }
+              end
+            end
+          rescue => e
+            log_error("Failed to scan orphaned message queues: #{e.message}")
+            EventMachine.next_tick { callback.call }
+          end
+        end
+
+        scan_batch.call("0")
+      rescue => e
+        log_error("Failed to cleanup orphaned message queues: #{e.message}")
+        EventMachine.next_tick { callback.call }
+      end
+
+      # Delete message queues in batches
+      def cleanup_message_queues_batched(queue_keys, batch_size, &callback)
+        return EventMachine.next_tick { callback.call } if queue_keys.empty?
+
+        total = queue_keys.size
+        batches = queue_keys.each_slice(batch_size).to_a
+
+        process_batch = lambda do |batch_index|
+          if batch_index >= batches.size
+            puts "[Faye::Redis::SubscriptionManager] INFO: Cleaned up #{total} orphaned message queues" if @options[:log_level] != :silent
+            EventMachine.next_tick { callback.call }
+            return
+          end
+
+          batch = batches[batch_index]
+
+          @connection.with_redis do |redis|
+            redis.pipelined do |pipeline|
+              batch.each { |key| pipeline.del(key) }
+            end
+          end
+
+          # Yield control between batches
+          EventMachine.next_tick { process_batch.call(batch_index + 1) }
+        end
+
+        process_batch.call(0)
+      rescue => e
+        log_error("Failed to cleanup message queues batch: #{e.message}")
+        EventMachine.next_tick { callback.call }
+      end
+
       # Async version of cleanup_empty_channels that yields between operations
       def cleanup_empty_channels_async(namespace, &callback)
         @connection.with_redis do |redis|

data/lib/faye/redis/version.rb

@@ -1,5 +1,5 @@
 module Faye
   class Redis
-    VERSION = '1.0.9'
+    VERSION = '1.0.10'
   end
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: faye-redis-ng
 version: !ruby/object:Gem::Version
-  version: 1.0.9
+  version: 1.0.10
 platform: ruby
 authors:
 - Zac