ワンクリックで
volcano-scheduler-config
View Volcano scheduler configuration. Check scheduler ConfigMap, actions, plugins, and tier settings.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
メニュー
View Volcano scheduler configuration. Check scheduler ConfigMap, actions, plugins, and tier settings.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Show and ping the gateway of a network interface, on a Kubernetes node or inside a pod's network namespace. Auto-detects the gateway from the routing table (ip -j route), reports interface type (RoCE / Ethernet / IB), and tests reachability with ping. Use for default-route / gateway questions, network reachability checks, RoCE/RDMA data-path validation, and "can this node/pod reach its gateway" investigations.
Retrieve logs from a Kubernetes node. Supports journalctl (systemd units) and file-based logs. Use when you need to inspect node-level logs (containerd, kubelet, etc.). Run via host_script (preferred) or node_script.
Guide for writing and improving Siclaw skills. Read this when creating or modifying a skill. Covers skill directory layout, SKILL.md format, script execution modes, and best practices.
Check node health and diagnose node-level issues (NotReady, DiskPressure, MemoryPressure, PIDPressure). Inspects node conditions, resource allocation, and real-time usage.
Diagnose DNS resolution failures in the cluster (NXDOMAIN, timeouts, SERVFAIL). Checks CoreDNS health, service endpoints, and DNS configuration.
Diagnose NetworkPolicy-related connectivity issues (traffic unexpectedly blocked, default-deny effects, egress blocking DNS). Identifies which NetworkPolicies affect a pod, checks ingress/egress rules, and verifies CNI support.
| name | volcano-scheduler-config |
| description | View Volcano scheduler configuration. Check scheduler ConfigMap, actions, plugins, and tier settings. |
View Volcano scheduler configuration to understand scheduling policies, enabled plugins, and actions. This skill helps diagnose configuration-related scheduling behaviors.
Scope: This skill is for diagnosis only. It retrieves configuration for analysis but does not modify any cluster state.
bash skills/core/volcano-scheduler-config/scripts/get-scheduler-config.sh [options]
| Parameter | Required | Description |
|---|---|---|
--section SECTION | no | Show specific section: actions, plugins, tiers, all (default: all) |
--format FORMAT | no | Output format: yaml, json, summary (default: summary) |
--raw | no | Show raw ConfigMap data without parsing |
Get summary of scheduler configuration:
bash skills/core/volcano-scheduler-config/scripts/get-scheduler-config.sh
View actions configuration only:
bash skills/core/volcano-scheduler-config/scripts/get-scheduler-config.sh --section actions
View plugins configuration:
bash skills/core/volcano-scheduler-config/scripts/get-scheduler-config.sh --section plugins
Show full YAML configuration:
bash skills/core/volcano-scheduler-config/scripts/get-scheduler-config.sh --format yaml
Get raw ConfigMap:
bash skills/core/volcano-scheduler-config/scripts/get-scheduler-config.sh --raw
The Volcano scheduler configuration is stored in:
volcano-systemvolcano-scheduler-configmapvolcano-scheduler.confactions: "enqueue, allocate, backfill" # Pipeline order
tiers:
- plugins:
- name: priority
- name: gang
- name: conformance
- plugins:
- name: overcommit
- name: drf
- name: predicates
- name: proportion
- name: nodeorder
- name: binpack
Actions define the two-phase scheduling pipeline:
A job can pass enqueue (PodGroup moves to Inqueue) but fail allocation (pods stay Pending) if node-level constraints block placement. This two-phase model explains the common scenario: "queue has capacity but pods won't schedule."
Actions in order:
| Action | Required | Purpose |
|---|---|---|
enqueue | Yes | Admit pod groups to queue |
allocate | Yes | Allocate resources to pods |
backfill | No | Fill idle resources with best-effort pods |
preempt | No | Evict low-priority pods for high-priority |
reclaim | No | Reclaim resources from over-allocated queues |
elect | No | Select target workload (removed in v1.6+) |
Tiers divide plugins into priority levels:
Common tier organization:
| Plugin | Function | Critical For |
|---|---|---|
gang | Gang scheduling | Batch jobs, ML training |
priority | Priority sorting | Workload prioritization |
conformance | Protect critical pods | System stability |
| Plugin | Function | Use Case |
|---|---|---|
proportion | Fair share allocation | Multi-tenant clusters |
drf | Dominant Resource Fairness | Fair GPU/CPU sharing |
overcommit | Allow overcommit | Resource efficiency |
| Plugin | Function | Use Case |
|---|---|---|
predicates | Node filtering | Resource/affinity matching |
nodeorder | Node ranking | Node selection optimization |
binpack | Dense packing | Reduce fragmentation |
numaaware | NUMA topology | HPC workloads |
| Plugin | Function | Use Case |
|---|---|---|
priority | Queue ordering | Queue prioritization |
proportion | Proportional share | Fair queue allocation |
bash skills/core/volcano-scheduler-config/scripts/get-scheduler-config.sh --section plugins
Look for name: gang in the plugin list. If missing, Gang scheduling will not work.
bash skills/core/volcano-scheduler-config/scripts/get-scheduler-config.sh --section actions
Look for reclaim in the actions list. If missing, queues cannot reclaim resources.
bash skills/core/volcano-scheduler-config/scripts/get-scheduler-config.sh --section plugins
Look for name: proportion. If missing, queue fair-share allocation is disabled.
Default configuration:
actions: "enqueue, allocate, backfill"
tiers:
- plugins:
- name: priority
- name: gang
enablePreemptable: false
- name: conformance
- plugins:
- name: overcommit
- name: drf
enablePreemptable: false
- name: predicates
- name: proportion
- name: nodeorder
- name: binpack
If your configuration differs significantly, it may explain scheduling behaviors.
Some plugins accept arguments:
- name: overcommit
arguments:
overcommit-factor: "1.2"
Use --format yaml or --raw to see full plugin configurations including arguments.
Symptom: PodGroups stay Pending, minMember never satisfied
Check:
bash skills/core/volcano-scheduler-config/scripts/get-scheduler-config.sh --section plugins | grep gang
Solution: Add gang plugin to configuration
Symptom: Queues cannot reclaim resources from over-allocated queues
Check:
bash skills/core/volcano-scheduler-config/scripts/get-scheduler-config.sh --section actions | grep reclaim
Solution: Add reclaim to actions list
Symptom: Unexpected scheduling behavior
Check:
bash skills/core/volcano-scheduler-config/scripts/get-scheduler-config.sh --section actions
Common orders:
enqueue, allocate, backfillenqueue, allocate, backfill, preemptenqueue, allocate, backfill, reclaimenqueue, allocate, backfill, preempt, reclaimImportant: enqueue must come before allocate. allocate must be present.
Symptom: Priority/preemption not working as expected
Check:
bash skills/core/volcano-scheduler-config/scripts/get-scheduler-config.sh --section tiers
Guideline:
| Plugin | Argument | Default | Description |
|---|---|---|---|
gang | enablePreemptable | true | Allow Gang pods to be preempted |
overcommit | overcommit-factor | 1.2 | Multiplier for allocatable resources |
drf | enablePreemptable | true | Allow DRF pods to be preempted |
nodeorder | various weights | 0 | Node scoring weights |
proportion | (none) | - | No arguments |
predicates | GPUSharingEnable | false | Enable GPU sharing |
jobEnqueueableFn from pluginsovercommit-factor of 1.2 means 20% overcommitallocate action for each tierpriority plugin for comparisonpreemptableFn from pluginsproportion pluginreclaimableFn from plugins| Variable | Default | Description |
|---|---|---|
VOLCANO_SCHEDULER_NS | volcano-system | Scheduler namespace |
VOLCANO_SCHEDULER_CONFIG | volcano-scheduler-configmap | ConfigMap name |
Human-readable summary:
Scheduler Configuration
=======================
Actions: enqueue, allocate, backfill
Tier 1 Plugins:
- priority
- gang
- conformance
Tier 2 Plugins:
- overcommit
- drf
- predicates
- proportion
- nodeorder
- binpack
Parsed YAML structure:
actions: "enqueue, allocate, backfill"
tiers:
- plugins:
- name: priority
- name: gang
Machine-parseable:
{
"actions": "enqueue, allocate, backfill",
"tiers": [
{
"plugins": [
{"name": "priority"},
{"name": "gang"}
]
}
]
}
ConfigMap raw output:
volcano-scheduler.conf:
---
actions: "enqueue, allocate, backfill"
tiers:
...
volcano-queue-diagnose - Queue resource analysisvolcano-gang-scheduling - Gang scheduling issuesvolcano-diagnose-pod - Pod scheduling diagnosisvolcano-scheduler-logs - Scheduler decision logs