Skip to content

GPUaaS Review Portal

Runbook index

Runbook index (full)¶

Runbook

All 38 runbooks under doc/operations/runbooks/, alphabetised. For the categorised view see Runbook inventory in /now.

A–C¶

Runbook
`Admin_Ops_Dashboard_Usage_Runbook.md`
`Agent_Orchestrator_v2_Runbook.md`
`API_Degradation_Runbook.md`
`App_Artifact_Lifecycle_Incident_Runbook.md`
`App_Catalog_Incident_Runbook.md`
`App_Platform_Operator_Incident_Runbook.md`
`App_Runtime_Billing_Incident_Runbook.md`
`App_Runtime_Lifecycle_Incident_Runbook.md`
`Billing_Worker_Failure_Runbook.md`
`CLI_Incident_and_Support_Triage_Runbook.md`

D–I¶

Runbook
`Database_Latency_or_Failover_Runbook.md`
`Enterprise_Federation_Incident_Runbook.md`
`Fleet_Telemetry_Incident_Runbook.md`
`GPU_Slice_Cleanup_Blocked_Slot_Runbook.md`
`GPU_Slice_Image_Pipeline_Runbook.md`
`GPU_Slice_Infra_Enablement_Proposal_v1.md`
`GPU_Slice_Node_Manual_Bootstrap_Runbook.md`
`IAM_Role_Assignment_and_Membership_Incident_Runbook.md`
`Incident_Communication_Runbook.md`

J–P¶

Runbook
`JWKS_Compromise_Breakglass_Runbook.md`
`Key_Rotation_and_Compromise_Response_Runbook.md`
`MAAS_H200_Host_Image_Pipeline_Runbook.md`
`Node_Agent_Control_Plane_Recovery_2026-03.md`
`Node_Onboarding_Runbook.md`
`Platform_Control_Disk_Cleanup_Runbook.md`
`Platform_Control_K3s_Recovery_Runbook.md`
`Provisioning_Workflow_Stuck_Runbook.md`
`Proxied_App_UI_Incident_Runbook.md`
`Python_SDK_Incident_and_Observability_Runbook.md`

Q–Z¶

Runbook
`Queue_Backlog_Runbook.md`
`Slurm_Reference_Deploying_Stuck_Runbook.md`
`Tenant_Project_Authorization_Runbook.md`
`Terminal_Gateway_Incident_Runbook.md`
`Three_Host_Lab_Incident_Runbook.md`
`User_Onboarding_Auth_Context_Runbook.md`
`Vault_Bootstrap_and_Root_Token_Runbook.md`
`Webhook_Processing_Outage_Runbook.md`

Catalog manifest¶

doc/operations/runbooks/runbooks.catalog.json is the machine-readable index used by the in-product admin runbook panel. New runbooks must register there before being linked from alerts.

Format requirements¶

Every runbook contains:

When to use — alert/symptom that opens it.
Owning team.
Diagnostic steps — ordered.
Mitigations — bounded actions.
Escalation — when to open an RCA / page senior on-call.
Evidence to capture — timestamps, queries, links.

Where to look next¶