一键导入
darwin-k8s-mcp
Queue-state and pipeline condition interpretation for remote build clusters
用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
菜单
Queue-state and pipeline condition interpretation for remote build clusters
用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
| name | darwin-k8s-mcp |
| description | Queue-state and pipeline condition interpretation for remote build clusters |
| roles | ["sysadmin","developer"] |
| modes | ["investigate","execute","plan","review","analyze"] |
Kueue Workloads are namespace-scoped CRDs (workloads.kueue.x-k8s.io) that
track queue admission state for pipeline runs on build clusters.
The Workload Admitted condition is the definitive queue signal:
| Condition | Status | Meaning |
|---|---|---|
Admitted | True | Workload admitted -- pipeline is executing |
Admitted | False | Workload queued -- waiting for capacity |
| (absent) | -- | Workload newly created or Kueue not yet reconciled -- treat as not admitted |
Workloads correlate to PipelineRuns via owner references. Prefer owner references over name-pattern matching for correlation.
A PipelineRun gated by Kueue shows reason: PipelineRunPending in
.status.conditions[type=Succeeded]. The tekton-kueue controller sets
spec.status: PipelineRunPending on creation; Kueue clears it on admission.
| Reason | Meaning |
|---|---|
PipelineRunPending | Pipeline is queued -- not started, waiting for admission |
Running | Pipeline is actively executing tasks |
Succeeded / Failed | Terminal state |
When a PipelineRun shows PipelineRunPending, it has not started. Report
queue state rather than deferring on assumed execution time.
The Workload Admitted condition is the source of truth for queue state.
PipelineRun conditions reflect the consequence of admission decisions but
may lag behind. If the Workload shows Admitted: True but the PipelineRun
still shows PipelineRunPending, the pipeline is transitioning -- it will
start shortly.
Do not re-check immediately on signal disagreement. The Workload condition is authoritative; report what it says.
tekton-kueue controller has not yet
created the Workload. This is normal immediately after pipeline creation.
Report "Workload not yet created" and rely on PipelineRun conditions.Admitted condition: Kueue has not yet
reconciled. Treat as not admitted (queued).Namespace-scoped resources (PipelineRuns, TaskRuns, Workloads, pods) are readable on remote clusters. Cluster-scoped resources (ClusterQueues) are not accessible.
Check PipelineRun and Workload conditions when a pipeline has been in a
non-terminal state longer than expected -- not immediately after triggering
a pipeline. The tekton-kueue controller and Kueue need time to create and
reconcile the Workload after pipeline creation.
Safety rules for modifying Dockerfiles and Containerfiles. Use when editing container build files.
MR/PR context gathering and diagnostic reporting guidelines. Activates when working on events that reference MRs/PRs or when reporting investigation findings.
Team coordination for implement and test modes. Developer and QE use team_send_results for final reports, team_huddle for mid-task questions to FRIDAY.
GitOps workflow rules for modifying infrastructure via git. Use when cloning repos, modifying Helm values, committing, pushing, or verifying deployments.
MR/PR lifecycle operations -- pipeline check, retest, merge, conflict reporting. Extends darwin-gitlab-ops.
Report findings and status updates to the Darwin FRIDAY. Use team_send_results for final reports (task modes). In message mode, use team_send_message instead.