Veer15/openenv-distributed-systems-debugging

1name: distributed-systems-debug-env2version: "1.0.0"3description: >4  An RL environment for debugging a distributed job processing pipeline with5  five deterministic fault scenarios: cascading timeout, byzantine queue fault,6  distributed lock starvation, backpressure cascade, and route partition.7author: Team Rocket8tags:9  - openenv10  - distributed-systems11  - debugging12  - reinforcement-learning13 14reward_range: [0.0, 1.0]15 16tasks:17  - name: cascading-timeout18    description: Synchronous upstream delay exceeds gateway timeout.19    difficulty: easy20    max_steps: 1521 22  - name: byzantine-queue-fault23    description: Poison pill message causes worker parse crash-loop.24    difficulty: medium25    max_steps: 1826 27  - name: distributed-lock-starvation28    description: Stale distributed lock prevents queue consumption.29    difficulty: hard30    max_steps: 2031 32  - name: backpressure-cascade33    description: Consumer throughput lower than producer enqueue rate.34    difficulty: hard35    max_steps: 2036 37  - name: route-partition38    description: Route policy blocks gateway to redis communication.39    difficulty: hard40    max_steps: 2041 42  - name: registry-corruption43    description: Gateway reloads a corrupted auth registry entry and fails requests.44    difficulty: medium45    max_steps: 1846 47  - name: job-generator-runaway48    description: Runaway enqueue rate overwhelms worker throughput and grows backlog.49    difficulty: hard50    max_steps: 2051 52observation_space:53  type: object54  properties:55    command_output:56      type: string57    metrics:58      type: object59      properties:60        gateway_success_rate:61          type: number62          minimum: 0.063          maximum: 1.064        gateway_p99_latency_ms:65          type: number66          minimum: 0.067        queue_depth:68          type: integer69          minimum: 070        worker_restart_count:71          type: integer72          minimum: 073        consumer_stall_count:74          type: integer75          minimum: 076    process_status:77      type: object78      additionalProperties:79        type: string80 81action_space:82  type: object83  properties:84    command:85      type: string86      description: Single bash command executed in the debug sandbox.87