---
name: pd-tso-or-scheduling
description: Diagnose PD, TSO, scheduling, placement, region-health, and ETCD operational root causes.
---

# Use When

Use this family subskill when the primary mechanism is PD/TSO, placement rules,
scheduling behavior, region health, ETCD capacity, PD leader pressure, or PD operational
failure.

# Required Evidence

- TSO wait/latency, PD leader CPU, scheduler, region checker, placement rule, region heartbeat,
  ETCD capacity, or PD log evidence.
- Recent operation context such as placement-rule change, scatter, evict-leader, TLS, scale,
  or dashboard/prometheus scrape issue.

# Classification

- `standard_root_cause_class`: `pd_tso_or_scheduling`
- compatible `primary_class`: `service_unavailability_or_worker_failure` until schema is extended
- common symptom tags: `pd_tso_wait`, `capacity_pressure`, `network_latency`
- specialist skills: `pd-oncall-handbook`

# Decision Rules

- Do not promote PD logs to primary solely because they overlap the incident window.
- Decide whether the risk is loss of TSO service, degraded scheduling, placement issue, ETCD
  capacity, or dashboard-only behavior.
- Use the precise standard class even when the compatible schema class is broad.

# Final Response Anchors

- Name the PD mechanism: TSO, scheduling, placement, region health, ETCD, or operational path.
- State whether the impact is control-plane only or data-plane visible.