| name | managing-k8s-loki-stack |
| description | Use when working with K8S Loki Stack — loki stack management for Kubernetes
log aggregation. Covers Loki deployment health, Promtail/log agent status, log
stream ingestion, storage backend configuration, retention policies, and query
performance. Use when debugging log ingestion issues, auditing logging
coverage, reviewing Loki storage, or troubleshooting Promtail scrape configs.
|
| connection_type | k8s |
| preload | false |
Loki Stack Management Skill
Manage and analyze Loki log aggregation stack including Loki, Promtail, and related components.
MANDATORY: Discovery-First Pattern
Always check Loki and log agent health before querying log streams.
Phase 1: Discovery
#!/bin/bash
echo "=== Loki Deployment ==="
kubectl get statefulset,deployment -l app.kubernetes.io/name=loki --all-namespaces -o custom-columns='NAMESPACE:.metadata.namespace,KIND:.kind,NAME:.metadata.name,READY:.status.readyReplicas,REPLICAS:.spec.replicas' 2>/dev/null
echo ""
echo "=== Loki Pods ==="
kubectl get pods -l app.kubernetes.io/name=loki --all-namespaces -o custom-columns='NAMESPACE:.metadata.namespace,NAME:.metadata.name,STATUS:.status.phase,RESTARTS:.status.containerStatuses[0].restartCount,NODE:.spec.nodeName' 2>/dev/null
echo ""
echo "=== Promtail DaemonSet ==="
kubectl get daemonset -l app.kubernetes.io/name=promtail --all-namespaces -o custom-columns='NAMESPACE:.metadata.namespace,NAME:.metadata.name,DESIRED:.status.desiredNumberScheduled,CURRENT:.status.currentNumberScheduled,READY:.status.numberReady' 2>/dev/null
echo ""
echo "=== Log Agents (Promtail/Alloy/FluentBit) ==="
kubectl get pods -l app.kubernetes.io/name=promtail --all-namespaces -o custom-columns='NAMESPACE:.metadata.namespace,NAME:.metadata.name,STATUS:.status.phase,NODE:.spec.nodeName' 2>/dev/null | head -15
kubectl get pods -l app.kubernetes.io/name=alloy --all-namespaces -o custom-columns='NAMESPACE:.metadata.namespace,NAME:.metadata.name,STATUS:.status.phase' 2>/dev/null | head -10
echo ""
echo "=== Loki Services ==="
kubectl get svc -l app.kubernetes.io/name=loki --all-namespaces -o custom-columns='NAMESPACE:.metadata.namespace,NAME:.metadata.name,TYPE:.spec.type,PORTS:.spec.ports[*].port' 2>/dev/null
Phase 2: Analysis
#!/bin/bash
echo "=== Loki Ring Health ==="
LOKI_NS=$(kubectl get pods -l app.kubernetes.io/name=loki --all-namespaces -o jsonpath='{.items[0].metadata.namespace}' 2>/dev/null)
LOKI_POD=$(kubectl get pods -l app.kubernetes.io/name=loki -n "$LOKI_NS" -o jsonpath='{.items[0].metadata.name}' 2>/dev/null)
kubectl exec -n "$LOKI_NS" "$LOKI_POD" -- wget -qO- http://localhost:3100/ready 2>/dev/null
echo ""
echo ""
echo "=== Loki Ingester Status ==="
kubectl exec -n "$LOKI_NS" "$LOKI_POD" -- wget -qO- http://localhost:3100/metrics 2>/dev/null | grep -E "loki_ingester_streams_created_total|loki_distributor_lines_received_total" | head -5
echo ""
echo "=== Promtail Logs (errors) ==="
PROMTAIL_NS=$(kubectl get daemonset -l app.kubernetes.io/name=promtail --all-namespaces -o jsonpath='{.items[0].metadata.namespace}' 2>/dev/null)
kubectl logs daemonset/promtail -n "$PROMTAIL_NS" --tail=20 2>/dev/null | grep -i "error\|fail\|drop" | head -10
echo ""
echo "=== Loki Logs (errors) ==="
kubectl logs -l app.kubernetes.io/name=loki -n "$LOKI_NS" --tail=20 2>/dev/null | grep -i "error\|fail\|warn" | head -10
kubectl get pvc -l app.kubernetes.io/name=loki --all-namespaces -o custom-columns= 2>/dev/null
PROMTAIL_DESIRED=$(kubectl get daemonset -l app.kubernetes.io/name=promtail --all-namespaces -o jsonpath= 2>/dev/null)
PROMTAIL_READY=$(kubectl get daemonset -l app.kubernetes.io/name=promtail --all-namespaces -o jsonpath= 2>/dev/null)
Output Format
- Target ≤50 lines per output
- Use
-o custom-columns for resource listings
- Show DaemonSet coverage (desired vs ready) for log agents
- Aggregate log volume metrics when available
- Never dump full Loki configs -- show storage backend and retention only
Anti-Hallucination Rules
- NEVER assume resource names — always discover via CLI/API in Phase 1 before referencing in Phase 2.
- NEVER fabricate metric names or dimensions — verify against the service documentation or
--help output.
- NEVER mix CLI commands between service versions — confirm which version/API you are targeting.
- ALWAYS use the discovery → verify → analyze chain — every resource referenced must have been discovered first.
- ALWAYS handle empty results gracefully — an empty response is valid data, not an error to retry.
Counter-Rationalizations
| Shortcut | Counter | Why |
|---|
| "I'll skip discovery and check known resources" | Always run Phase 1 discovery first | Resource names change, new resources appear — assumed names cause errors |
| "The user only asked for a quick check" | Follow the full discovery → analysis flow | Quick checks miss critical issues; structured analysis catches silent failures |
| "Default configuration is probably fine" | Audit configuration explicitly | Defaults often leave logging, security, and optimization features disabled |
| "Metrics aren't needed for this" | Always check relevant metrics when available | API/CLI responses show current state; metrics reveal trends and intermittent issues |
| "I don't have access to that" | Try the command and report the actual error | Assumed permission failures prevent useful investigation; actual errors are informative |
Common Pitfalls
- Deployment modes: Loki runs as monolithic, simple-scalable, or microservices -- check architecture before debugging
- Promtail coverage: DaemonSet must run on all nodes -- check tolerations for control-plane nodes
- Storage backend: filesystem (single instance), S3/GCS/Azure (scalable) -- check
storage_config in ConfigMap
- Retention: Set via
compactor.retention_enabled and limits_config.retention_period -- not enabled by default
- Rate limiting: Loki enforces per-tenant rate limits -- check for
rate limit errors in distributor logs
- Chunk encoding: snappy is default -- gzip saves space but uses more CPU
- Label cardinality: High-cardinality labels cause performance issues -- avoid dynamic labels like pod names
- Log level: Promtail pipeline stages can filter/transform logs -- check
pipelineStages in config
