config/jobmgr/base.yaml

storage:
  cassandra:
    max_parallel_batches: 1000
    max_updates_job: 10
    connection:
      contactPoints: ["127.0.0.1"]
      port: 9042
      consistency: LOCAL_QUORUM
      serialConsistency: LOCAL_SERIAL
      hostPolicy: TokenAwareHostPolicy
      # Need to increase timeout from 10s to 20s to avoid recovery code from timing out
      # We saw recovery code timing out when peloton was recovering from a
      # Cassandra latency spike issue.
      timeout: 20s
    store_name: peloton_test
    migrations: pkg/storage/cassandra/migrations/
  use_cassandra: false
  db_write_concurrency: 40

job_manager:
  http_port: 5292
  grpc_port: 5392
  goal_state:
    job_batch_runtime_update_interval: 10s
    job_service_runtime_update_interval: 1s
  task_launcher:
    placement_dequeue_limit: 10
    get_placements_timeout_ms: 100
  task_evictor:
    eviction_period: 60s
    eviction_dequeue_limit: 100
    eviction_dequeue_timeout_ms: 100
  deadline:
    deadline_tracking_period: 30m
  job_service:
    # TODO (adityacb): Adjust this limit once we fix T1689063 and T1689077
    # and have a better data model
    max_tasks_per_job: 100000
    med_instance_count: 500
    high_instance_count: 1000
    low_get_workflow_events_workers: 25
    med_get_workflow_events_workers: 50
    high_get_workflow_events_workers: 100
    enable_secrets: false
    thermos_executor:
      path: "/usr/share/aurora/bin/thermos_executor.pex"
      flags: "--preserve_env --nosetuid-health-checks --nosetuid --no-create-user"
  # Refresh AciveTaskCache every 5 min
  active_task_update_period: 300s
  # being deprecated
  job_runtime_calculation_via_cache: false
  workflow_progress_check:
    # check all the workflow progress every 30 min
    workflow_progress_check_period: 30m
    # if a workflow is not updated for 30min,
    # consider it to be stale
    stale_workflow_threshold: 30m

election:
  root: "/peloton"

health:
  heartbeat_interval: 5s

metrics:
  runtime_metrics:
    enabled: true
    interval: 10s

rate_limit:
  enabled: false
  methods:
    # order of the entries matters here.
    # a procedure will try to match each entry from top to bottom,
    # and the first satisfied config will be used.
  - name: 'peloton.api.v1alpha.job.stateless.svc.JobService:GetJob'
    rate: 100
    burst: 100
  - name: 'peloton.api.v1alpha.job.stateless.svc.JobService:Get*'
    rate: 100
    burst: 100
  - name: 'peloton.api.v1alpha.job.stateless.svc.JobService:Browse*'
    rate: 100
    burst: 100
  - name: 'peloton.api.v1alpha.job.stateless.svc.JobService:List*'
    rate: -1 # if either burst or rate is < 0, no rate limit

#  # if a user calls a method not listed in methods list,
#  # it would fall through to this branch
#  default:
#    rate: -1
#    burst: -1

# TODO: need to find a way to auto generate the list
api_lock:
  read_apis:
    - '*:Get*'
    - '*:Query*'
    - '*:List*'
    - '*:Browse*'
    - 'peloton.api.v1alpha.watch.svc.WatchService:*'
  write_apis:
    - '*:Create*'
    - '*:Delete*'
    - '*:Start*'
    - '*:Stop*'
    - '*:Restart*'
    - '*:Refresh*'
    - '*:Update*'
    - '*:Pause*'
    - '*:Resume*'
    - '*:Rollback*'
    - '*:Abort*'
    - '*:Replace*'
    - '*:Patch*'