rolling-update

Rolling Update with Batch Execution and Automatic Rollback

Updating all replicas at once risks a full outage if the new version has a bug. This workflow analyzes the current deployment, plans a rolling strategy (1 replica at a time, 20% max unavailable), executes the update in batches, and verifies all replicas are healthy -- with automatic rollback if any batch fails.

Workflow

service, newVersion
       |
       v
+--------------+     +---------------+     +------------------+     +-----------------+
| ru_analyze   | --> | ru_plan       | --> | ru_execute       | --> | ru_verify       |
+--------------+     +---------------+     +------------------+     +-----------------+
  ANALYZE-1342        strategy: rolling     batchesCompleted         allHealthy=true
  currentReplicas     batchSize: 1          = totalBatches           rollbackOccurred
  currentVersion      maxUnavailable: 20%   rollbackTriggered        = false
  healthyReplicas     rollbackOnFailure     = false

Workers

Analyze -- Takes service and newVersion. Returns analyzeId: "ANALYZE-1342" with currentReplicas, currentVersion, and healthyReplicas.

PlanUpdate -- Plans the rolling strategy: batchSize: 1, maxUnavailable: max(1, replicas/5), totalBatches equal to replica count, and rollbackOnFailure: true.

ExecuteUpdate -- Executes the rolling update across all batches. Returns batchesCompleted equal to totalBatches, rollbackTriggered: false, and an updatedAt timestamp.

VerifyUpdate -- Confirms all replicas are healthy. allHealthy is true when no rollback was triggered. Returns rollbackOccurred flag and a completedAt timestamp.

Tests

33 unit tests cover deployment analysis, update planning, batch execution, and health verification.

Running

See ../../RUNNING.md for setup and execution instructions.