check-quota

When to Use This Skill

Run a quota check before:

• Creating a VM (CPU + memory + storage consumption)
• Deploying an application (CPU + memory + pods)
• Creating a managed Kubernetes cluster (large CPU + memory + storage)
• Provisioning a database (storage + pods)
• Allocating a public EIP (publicIPv4 limit)

Also use for troubleshooting when workloads fail with exceeded quota errors.

---

Organization-Level Quota

The Organization resource exposes aggregated usage across all projects in .status.quotaUsage. Values are refreshed every 5–7 minutes by the platform controller.

kubectl get organization {org} -n {org} \
  -o jsonpath='{.status.quotaUsage}' | jq .

Expected output:

{
  "cpu":           { "used": "18.975", "hard": "26" },
  "memory":        { "used": "63.6Gi", "hard": "70Gi" },
  "storage":       { "used": "443.2Gi","hard": "460Gi" },
  "pods":          { "used": "33",     "hard": "500" },
  "publicIPv4":    { "used": "3",      "hard": "3" },
  "objectStorage": { "used": "",       "hard": "500Gi" },
  "lastUpdated":   "2026-04-07T20:55:42Z"
}

Field reference:

• cpu — cores (decimal). e.g. "18.975" = 18,975 millicores
• memory / storage — GiB consumed vs plan hard limit
• publicIPv4 — count of externalNetworkType: public EIPs across all org namespaces
• objectStorage — hard limit from plan; used is populated asynchronously
• lastUpdated — timestamp of last controller refresh (up to 7 min old)

Check a single field:

# CPU remaining
kubectl get organization {org} -n {org} \
  -o jsonpath='{.status.quotaUsage.cpu}' | jq .
# → { "hard": "26", "used": "18.975" }

---

Project-Level Quota

Each Project resource exposes per-namespace usage in .status.quotaUsage:

kubectl get project {project} -n {org} \
  -o jsonpath='{.status.quotaUsage}' | jq .

Expected output:

{
  "cpu":               { "used": "6.72",    "hard": "26" },
  "memory":            { "used": "16.824Gi","hard": "70Gi" },
  "storage":           { "used": "147.4Gi", "hard": "460Gi" },
  "pods":              { "used": "12",      "hard": "500" },
  "perProjectQuotaSet": false,
  "lastUpdated":       "2026-04-07T20:55:00Z"
}

• hard shows the org-wide limit when perProjectQuotaSet: false, or the per-project cap when set by an admin
• perProjectQuotaSet: true means this project has an explicit ResourceQuota/project-quota that may be tighter than the org limit

All projects at a glance:

kubectl get projects -n {org} \
  -o custom-columns='PROJECT:.metadata.name,CPU_USED:.status.quotaUsage.cpu.used,CPU_HARD:.status.quotaUsage.cpu.hard,MEM_USED:.status.quotaUsage.memory.used,MEM_HARD:.status.quotaUsage.memory.hard'

---

Real-Time Enforcement State

quotaUsage is refreshed every 5–7 min. For the live enforcement state (what Kubernetes is actively enforcing right now), query the underlying ResourceQuota objects:

# All quotas in a project namespace
kubectl get resourcequota -n {org}-{project}

# Detailed breakdown with usage bars
kubectl describe resourcequota -n {org}-{project}

The hrq.hnc.x-k8s.io quota is the organization-wide HNC propagated limit. The project-quota quota (if present) is the per-project cap set by an admin.

Note: kubectl describe resourcequota shows raw Kubernetes units — millicores for CPU (e.g. 6720m) and bytes for memory/storage (e.g. 18064129473). Use .status.quotaUsage on the Project or Organization resource for human-readable values.

---

Interpreting Results

Sufficient capacity

cpu:     used=6.72  / hard=26    → 19.28 cores free ✅
memory:  used=16Gi  / hard=70Gi  → 54Gi free        ✅
storage: used=147Gi / hard=460Gi → 313Gi free       ✅

Proceed with the deployment.

Near limit — warn user

Any field where used / hard > 0.8 (80%) is worth flagging before proceeding with large workloads.

At or over limit — action required

publicIPv4: used=3 / hard=3  → 0 free ❌

Options:

• Free existing resources: delete unused EIPs, VMs, or pods
• Add a Turbo Add-on: navigate to Manage Organization → Billing → Turbo Add-ons
• Upgrade plan: for publicIPv4, upgrade to Scale Pool (3 IPs) or higher

---

Troubleshooting: "exceeded quota" Errors

When a workload fails with:

exceeded quota: plan-quota, requested: requests.cpu=500m,
used: requests.cpu=25500m, limited: requests.cpu=26

1. Run org quota check to find the exhausted resource:

   kubectl get organization {org} -n {org} -o jsonpath='{.status.quotaUsage}' | jq .
   

2. Run project quota check to see which project is consuming the most:

   kubectl get projects -n {org} \
     -o custom-columns='PROJECT:.metadata.name,CPU:.status.quotaUsage.cpu.used,MEM:.status.quotaUsage.memory.used'
   

3. Free capacity or expand quota before retrying.

---

Quota Limits by Plan

Resource	Dev Pool	Pro Pool	Scale Pool
CPU (requests)	4 cores	8 cores	16 cores
Memory	8 Gi	24 Gi	56 Gi
Storage	60 Gi	160 Gi	320 Gi
Pods	100	200	500
Public IPv4	1	1	3
Object Storage	20 Gi	100 Gi	500 Gi

Turbo x1 adds: +2 CPU, +4 Gi RAM, +20 Gi storage (€9/mo). Turbo x2 adds: +4 CPU, +8 Gi RAM, +40 Gi storage (€16/mo).

---

Safety

• Always check quota before creating resource-heavy workloads (clusters, VMs with large disks)
• publicIPv4 quota is hard — no burst. Check before allocating any externalNetworkType: public EIP
• quotaUsage.lastUpdated may be up to 7 minutes stale; use kubectl describe resourcequota for real-time data when timing matters