composer require vortos/vortos-scheduler

php vortos vortos:migrate

use Vortos\Scheduler\DependencyInjection\VortosSchedulerConfig;

return static function (VortosSchedulerConfig $config): void {
    $config
        ->shardCount(4)
        ->leaseDriver('postgres-advisory')
        ->runRetentionDays(90);
};

# Lease backend. Options: sql (default), redis, postgres-advisory, in-memory
VORTOS_SCHEDULER_LEASE_DRIVER=sql

# Redis DSN for the redis lease driver (shares vortos-cache's DSN by default)
VORTOS_CACHE_DSN=redis://redis:6379

# Shard count for horizontal scaling of the fire loop (advanced — see Production)
SCHEDULER_SHARD_COUNT=1

# Daemon poll interval when no shard has due work, in seconds
SCHEDULER_MAX_IDLE_SEC=60

# Lease TTL in seconds. Must be longer than the slowest expected tick
SCHEDULER_LEASE_TTL_SEC=30

# Maximum age of missed slots to catch up, in seconds
# Slots older than this during misfire recovery are dropped
SCHEDULER_MAX_CATCHUP_AGE_SECONDS=86400

# Maximum concurrent fires per tenant before extras are skipped. 0 = no limit
SCHEDULER_TENANT_MAX_CONCURRENT_FIRES=0

# How long a dispatched job is assumed "done" when no completion event arrives, in seconds
SCHEDULER_ASSUMED_DONE_TTL_SECONDS=3600

# Circuit breaker: consecutive failures before the breaker opens
SCHEDULER_CB_FAILURE_THRESHOLD=5

# Circuit breaker: seconds to wait in the Open state before probing again
SCHEDULER_CB_RECOVERY_WINDOW_SEC=30

# In-process resolver cache TTL (seconds). Reduces store round-trips on CLI/admin paths
# The daemon itself bypasses the cache for zero-overhead hot path
SCHEDULER_RESOLVER_CACHE_TTL_SEC=5

# Dead-man detector: tolerance window before a metric alert fires, in seconds
SCHEDULER_DEADMAN_TOLERANCE_SEC=300

# Cardinality guard: maximum distinct schedule_id label values in Prometheus metrics
SCHEDULER_METRICS_MAX_CARDINALITY=200

# 4-eyes approval TTL, in seconds
SCHEDULER_APPROVAL_TTL_SECONDS=86400

# HMAC signing key for the audit hash-chain. Empty disables audit projection entirely
SCHEDULER_AUDIT_HMAC_KEY=

# Entries per audit HMAC checkpoint epoch
SCHEDULER_AUDIT_EPOCH_SIZE=1000

# --- Run retention (auto-prune) — see the Retention page for the full guide ---

# Days to keep completed/failed runs. 0 disables auto-prune entirely
SCHEDULER_RUN_RETENTION_DAYS=30

# Days to keep terminal (dispatched/failed) fire-queue rows. 0 disables. Pruned
# during the same daily sweep, so it needs SCHEDULER_RUN_RETENTION_DAYS > 0
SCHEDULER_FIRE_QUEUE_RETENTION_DAYS=7

# Rows deleted per chunk by the prune sweep (applies to both runs and fire-queue)
SCHEDULER_PRUNE_BATCH_SIZE=5000

# Wall-clock budget for one prune sweep
SCHEDULER_PRUNE_MAX_DURATION_SEC=240

# --- Fire-queue consumer (scheduler:consume) ---

# Rows claimed per consumer batch
SCHEDULER_CONSUME_BATCH_SIZE=50

# Sleep interval between empty polls in --loop mode
SCHEDULER_CONSUME_POLL_INTERVAL_SEC=2

# scheduler:doctor C11: how old the oldest pending fire-queue row may be before it's a Fail
SCHEDULER_CONSUME_STALL_THRESHOLD_SEC=120

# Requeue an unrunnable fire (unknown/incapable command class) this many times before dead-lettering it
SCHEDULER_FIRE_MAX_ATTEMPTS=10

# Exponential backoff base and cap (seconds) between fire requeues
SCHEDULER_FIRE_BACKOFF_BASE_SEC=2
SCHEDULER_FIRE_BACKOFF_CAP_SEC=300

VORTOS_SCHEDULER_LEASE_DRIVER=sql

VORTOS_SCHEDULER_LEASE_DRIVER=redis

VORTOS_SCHEDULER_LEASE_DRIVER=postgres-advisory

composer require vortos/vortos-scheduler-admin

php vortos vortos:assets:publish

php vortos scheduler:run
php vortos scheduler:consume --loop

php vortos scheduler:doctor