Development Workflow

Understanding stage categories helps compose valid pipelines. Stages must appear in this order:

Key ordering principle: Place $match as early as possible (reduces volume for all downstream stages). Place $project after $match (reduces document size). Place $https after windows (batches API calls).

5-Phase Development Lifecycle

Phase 1: Project Setup

Goal: Workspace and connections ready.

1. Discover existing resources:

   • atlas-streams-discover → list-workspaces — see what already exists
   • If workspace exists, inspect-workspace to review config

2. Create workspace (if needed):

   • atlas-streams-build → resource: "workspace"
   • Choose region close to your data sources
   • Start with tier: "SP10" for development
   • includeSampleData: true (default) gives you sample_stream_solar for testing

3. Verify workspace:

   • atlas-streams-discover → inspect-workspace — confirm state and region

Phase 2: Connection Development

Goal: All data sources and sinks connected and verified.

1. Identify required connections:

   • Source connections (Kafka, Cluster change streams, Kinesis, Sample)
   • Sink connections (Cluster for $merge, Kafka for $emit, S3, Kinesis)
   • Enrichment connections (Https for $https, Cluster for $lookup)

2. Create each connection:

   • atlas-streams-build → resource: "connection" for each
   • Let the tool elicit missing sensitive fields (passwords, bootstrap servers)
   • See connection-configs.md for type-specific schemas

3. Verify connections:

   • atlas-streams-discover → list-connections — confirm all created
   • atlas-streams-discover → inspect-connection for each — verify state and config

Phase 3: Processor Development

Goal: Working processor with validated pipeline.

Pre-Deployment Connection Validation (MANDATORY)

BEFORE creating any processor, you MUST validate all connections referenced in your pipeline. This prevents silent failures and confusion about data destinations.

Step 1: List all connections in workspace

Verify all required connections exist.

Step 2: Inspect EACH connection referenced in pipeline

For EVERY connectionName in your pipeline (source, sink, enrichment), inspect it:

Verify for each connection:

• Not done: Connection exists and state is READY
• Not done: Connection type matches intended usage:
  • Cluster: valid for $source (change streams), $merge, $lookup
  • Kafka: valid for $source, $emit
  • S3: valid for $emit only
  • Https: valid for $https enrichment or sink
  • Lambda: valid for $externalFunction only
• Not done: Connection name matches actual target (avoid confusion):
  • ⚠️ BAD: connection “atlascluster” → actual target “ClusterRestoreTest”
  • ✅ GOOD: connection “cluster-restore-test” → actual target “ClusterRestoreTest”
• Not done: For Cluster connections: verify the clusterName field points to the intended cluster

Step 3: Present validation summary to user

Always show the user what connections will be used:

Step 4: Wait for user confirmation if warnings exist

If any connection name doesn’t match its target, ask the user to confirm before proceeding.

Step 5: Only then create the processor

This validation workflow prevents:

• Creating processors with non-existent connections (fails immediately)
• Writing data to unexpected clusters (e.g., “atlascluster” → “ClusterRestoreTest” instead of “AtlasCluster”)
• Confusion when verifying output data later

Incremental Pipeline Development

Follow incremental pipeline development — test at each step:

Step 1: Basic connectivity

Create with autoStart: true. Verify documents flow. Stop processor.

Step 2: Add filtering

Modify pipeline (stop → modify-processor → start). Verify filtered output.

Step 3: Add transformations

Remember: $$NOW is NOT valid in streaming. Use a field from the source document or omit.

Step 4: Add windowing or enrichment (if needed)

Step 5: Add error handling

• Configure DLQ: {"dlq": {"connectionName": "my-sink", "db": "streams_dlq", "coll": "failed_docs"}}
• Add $ifNull for optional enrichment fields
• Set onError: "dlq" on $https stages

Phase 4: Testing & Validation

Goal: Processor verified working correctly.

1. Confirm processor state:

   • atlas-streams-discover → inspect-processor — state should be STARTED

2. Run diagnostics:

   • atlas-streams-discover → diagnose-processor — full health report

3. Verify data flow:

   • Use MongoDB count tool on output collection — documents arriving?
   • Use MongoDB find tool on output collection — data looks correct?
   • Use MongoDB count tool on DLQ collection — any errors?
   • If DLQ has documents, use MongoDB find tool to inspect failure reasons

4. Classify output volume:

   • See output-diagnostics.md for the full decision framework
   • Alert processors: low output is expected
   • Transformation processors: low output is a red flag

Phase 5: Production Deployment

Goal: Processor running at appropriate tier with monitoring.

1. Right-size the tier:

   • See sizing-and-parallelism.md for tier selection
   • Review memoryUsageBytes from diagnostics
   • Consider parallelism needs for $merge, $lookup, $https
   • Upgrade tier: atlas-streams-manage → stop-processor, then start-processor with tier override

2. Ensure DLQ is configured (mandatory for production)

3. Use descriptive processor names (e.g., fraud-detector, order-enricher, iot-rollup)

Debugging Decision Trees

Connection Failures

1. atlas-streams-discover → inspect-connection — check state
2. If Kafka: verify bootstrapServers is a comma-separated string (not array)
3. If Cluster: verify cluster exists in project (atlas-list-clusters)
4. If AWS (S3/Kinesis/Lambda): verify IAM role ARN is registered in Cloud Provider Access
5. If Https: verify URL is reachable and auth headers are in connection config

Processor Startup Failures

1. atlas-streams-discover → diagnose-processor — check state and errors
2. If FAILED: read the error message in diagnostics
3. Common causes:
   • Invalid pipeline syntax (missing $source, missing sink)
   • $$NOW/$$ROOT/$$CURRENT used (not valid in streaming)
   • Kafka $source missing topic field
   • Referenced connection doesn’t exist — validate with list-connections first
   • Connection name doesn’t match expected target — inspect connection to verify actual cluster/resource
   • OOM — tier too small for pipeline complexity

Processing Errors (Running but DLQ filling up)

1. Use MongoDB find tool on DLQ collection — inspect error messages
2. Common causes:
   • Schema mismatches in source data
   • $https enrichment failures (API down, auth expired)
   • Type errors in $addFields or $project expressions
3. Fix: stop-processor → modify-processor (fix pipeline) → start-processor

Performance Issues (Running but slow)

1. atlas-streams-discover → diagnose-processor — check stats
2. Check memoryUsageBytes — if near 80% of tier RAM, upgrade tier
3. Check if $match is early in pipeline (reduces downstream volume)
4. Check if $https has parallelism setting (increase for I/O-bound enrichment)
5. Check if windows have partitionIdleTimeout (idle Kafka partitions block windows)
6. Consider upgrading tier or increasing stage parallelism

Operational Monitoring Cadence

Daily

• Check processor states via atlas-streams-discover → list-processors
• Verify DLQ collections aren’t growing via MongoDB count tool
• Confirm output collections are receiving data

Weekly

• Run diagnose-processor for each production processor
• Review memoryUsageBytes trends — approaching 80%?
• Check connection health across all connections

Monthly

• Evaluate tier appropriateness — over-provisioned or under-provisioned?
• Review DLQ patterns — recurring errors that need pipeline fixes?
• Consider parallelism adjustments based on throughput trends

Troubleshooting

Pre-Deploy Quality Checklist

Before creating a processor, verify:

Connection Validation (MANDATORY - Always do this first)

• Not done: CRITICAL: Call atlas-streams-discover → action: "list-connections" to list all connections in workspace
• Not done: CRITICAL: Call atlas-streams-discover → action: "inspect-connection" for EACH connection referenced in pipeline
• Not done: CRITICAL: Verify connection names clearly indicate their actual targets (avoid generic names like “atlascluster” pointing to “ClusterRestoreTest”)
• Not done: CRITICAL: Present connection summary to user: “Connection ‘X’ → Actual target ‘Y’” for each connection
• Not done: CRITICAL: Warn user if connection names don’t match their targets and ask for confirmation
• Not done: All connections are in READY state
• Not done: Connection types match usage (Cluster for $source/$merge, Kafka for topics, etc.)

Pipeline Validation

• Not done: search-knowledge was called to validate sink/source field names
• Not done: Pipeline starts with $source and ends with $merge, $emit, $https, or $externalFunction (async)
• Not done: No $$NOW, $$ROOT, or $$CURRENT in the pipeline
• Not done: Kafka $source includes a topic field
• Not done: Kafka $source with windowed pipeline includes partitionIdleTimeout (prevents windows from stalling on idle partitions)
• Not done: HTTPS connections are only used in $https enrichment or sink stages, not in $source
• Not done: DLQ is configured (recommended for production)
• Not done: $https stages use onError: "dlq" (not "fail")
• Not done: $externalFunction stages use onError: "dlq" and execution is explicitly set
• Not done: API auth is stored in connection settings, not hardcoded in the pipeline

Post-Deploy Verification Workflow

After creating and starting a processor:

1. atlas-streams-discover → action: "inspect-processor" — confirm state is STARTED
2. atlas-streams-discover → action: "diagnose-processor" — check for errors in the health report
3. Use MongoDB count tool on the DLQ collection — verify no errors accumulating
4. Use MongoDB find tool on the output collection — verify documents are arriving
5. If output is low/zero, classify processor type before assuming a problem (see Debug section)