fix: AWS deployment system

.changeset/aws-sdk-migration.md

@@ -0,0 +1,37 @@
+---
+"pushkin-cli": major
+---
+
+Complete AWS Fargate deployment modernization and critical bug fixes
+
+This major update completely modernizes the AWS deployment infrastructure with critical bug fixes and improvements:
+
+**Breaking Changes:**
+- Migrated from AWS CLI to AWS SDK v3 for all AWS operations
+- Standardized database configuration to use `url` instead of `host` throughout
+
+**Critical Bug Fixes:**
+- Fix #364: Replace localhost with Service Discovery DNS for RabbitMQ in Fargate awsvpc networking mode
+- Fix #363: RabbitMQ passwords now persist across multiple deployments (idempotent behavior)
+
+**New Features:**
+- Added AWS Cloud Map (Service Discovery) support for ECS Fargate service-to-service communication
+- Improved DNS handling with subdomain fallback to parent hosted zones
+- Enhanced OAC (Origin Access Control) handling with retry logic
+- Added comprehensive CloudWatch logging for ECS tasks
+- Improved database connection reliability with exponential backoff retry logic
+
+**Improvements:**
+- Complete JSDoc documentation for all AWS-related functions
+- Better error messages and logging throughout deployment process
+- Standardized async/await patterns for better error handling
+- Improved security group and VPC configuration handling
+- Enhanced RDS instance management with proper wait conditions
+
+**Infrastructure Updates:**
+- Modernized all AWS service clients (S3, CloudFront, RDS, ECS, EC2, Route53, etc.)
+- Better handling of AWS resource lifecycle (create, update, delete)
+- Improved idempotency for AWS resource creation
+- Enhanced cleanup with aws armageddon command
+
+This update has been manually tested on AWS Fargate deployments and significantly improves the reliability and maintainability of AWS infrastructure management.

.changeset/fix-rabbitmq-heartbeat.md

@@ -0,0 +1,18 @@
+---
+"pushkin-worker": patch
+"@pushkin-templates/site-basic": patch
+---
+
+Fix RabbitMQ heartbeat timeout causing worker crashes
+
+**Bug Fix:**
+- Resolves "Heartbeat timeout" errors that prevented experiment workers from completing database operations
+- Workers would crash with "Error: Heartbeat timeout at Heart.<anonymous>" during experiment execution
+
+**Changes:**
+- Added `heartbeat: 30` configuration to `amqp.connect()` in pushkin-worker to send heartbeats every 30 seconds
+- Upgraded RabbitMQ from version 3.6 to 3.12 in docker-compose.dev.yml template
+- Added `RABBITMQ_HEARTBEAT: '30'` environment variable to RabbitMQ service configuration
+
+**Impact:**
+This fix ensures stable RabbitMQ connections during long-running experiment tasks and prevents connection timeouts that were blocking database persistence of user data and experiment results.

.gitignore

@@ -79,6 +79,8 @@ docker-compose.production.noEnvDependency.yml
 # next.js build output
 .next
 
+# yarn cache files
+**/.yarn
 # Docs build
 site/
 venv/
@@ -90,4 +92,9 @@ venv/
 test-results/
 playwright-report/
 blob-report/
-playwright/.cache/
+playwright/.cache/
+
+# CLAUDE files
+CLAUDE.local.md
+CLAUDE/*
+.claude

eslint.config.js

@@ -46,8 +46,8 @@ const monorepoConfig = {
           FunctionDeclaration: true,
           MethodDefinition: true,
           ClassDeclaration: true,
-          ArrowFunctionExpression: true,
-          FunctionExpression: true,
+          ArrowFunctionExpression: false,
+          FunctionExpression: false,
         },
       },
     ],