SignalForge
Op
Operator priorities
3 ranked actions
01

Restore payments-api capacity and stability first: investigate the CrashLoopBackOff pod, fix the failing replica, and get the Deployment back to 3/3 ready before making further changes.

02

Relieve cluster memory pressure and scheduling blockage: free memory or scale the AKS node pool so pending pods can schedule and the HPA has headroom beyond its current max replicas.

03

Reduce exposure of the payments namespace: confirm the LoadBalancer is required, then add namespace NetworkPolicies and tighten Service/ServiceAccount settings (automount token and seccomp) for the workload.

Kubernetes namespace
cluster:aks-prod-eu-1:namespace:payments
Hostname snapshot: aks-prod-eu-1 · Kubernetes (aks)
Artifact family
Kubernetes bundle
kubernetes-bundle
Source
upload
Mar 27, 02:24 AM
Artifact source
phase9b-e2e
Collector
signalforge-collectors
1.1.0
Target ID
cluster:aks-prod-eu-1:namespace:payments
Recorded at
Mar 27, 02:24 AM
Findings
16
7
high
9
medium
Run status
complete
Analysis completed successfully for this artifact snapshot.
Primary operator signal

Cluster Capacity Snapshot

Quantitative node-capacity signals and scheduling headroom captured directly from the Kubernetes bundle.

Needs action
Scope
Namespace payments
aks-prod-eu-1
Peak memory
91.0%
Low memory headroom
Peak CPU
92.0%
Scheduling warnings present
Node pressure
1
Nodes with NotReady or pressure conditions
Operator summary

Node Capacity Bars

Compact CPU and memory bars so operators can see which nodes are carrying the most pressure without reading raw metrics.

Watch closely
aks-system-000001 memory
14900Mi
91.0%
aks-system-000001 CPU
1850m
92.0%
Operator summary

Top Workload Consumers

Highest pod-level CPU and memory consumers captured by `kubectl top` during the run.

Watch closely
payments/payments-api-7f8d9
CPU 420m
530Mi
payments/payments-api-7f8d9
Memory 530Mi
420m
Operator summary

Cluster Guardrails

Autoscaling, disruption, quota, and namespace-default coverage that changes how operators should interpret capacity signals.

Watch closely
HPAs
1
Autoscaling objects present
Blocked PDBs
1
PDBs with zero allowed disruptions
Quota pressure
1
Quota resources at or above 90%
LimitRange coverage
0/1
Namespaces with default limits and requests
Pending claims
0
PersistentVolumeClaims still pending
Operator summary

Workload Instability

Short, operator-oriented callouts for scheduling, rollout, and failing-workload evidence.

Needs action
Scheduling pressure
0/3 nodes are available: 3 Insufficient memory.
Scheduling pressure
4 warning events captured in the bundle.
Deployment payments/payments-api
Ready 1/3, unavailable 2, observed generation 6 of 7.
Operator summary

Run Health Summary

A compact operator view of severity and signal distribution before you drop into detailed findings.

Needs action
Critical + high
7
Needs operator attention
Instability & pressure
11
Operational signal count
Identity & access
1
RBAC, tokens, service accounts, secrets
Exposure
2
Public reachability and listener posture
Findings table controls

Filter the findings table by signal or severity while keeping the current visible count in view.

16 of 16 visible·All signal buckets·All severities
Filter by signal
Filter by severity

Detailed review

Findings

16 findings
Analysis narrative
Full narrative summary

Expanded explanation for operators who want the model summary after reviewing the findings table.

▼
  • The payments namespace shows multiple concurrent reliability and availability issues: the main Deployment is not fully rolled out, pods are crashing, and the HPA is already maxed out while CPU is high.
  • Cluster capacity is under strain: node MemoryPressure is present, node CPU/memory are elevated, and scheduling events report insufficient memory.
  • The namespace is externally reachable via a LoadBalancer Service and also lacks a NetworkPolicy, increasing exposure of the payments API surface.
  • Safety controls are partially in place (non-root, read-only root filesystem, no privilege escalation), but pod security hardening is incomplete because seccomp is not set and service account tokens are auto-mounted.
  • Resource governance is tight: the namespace quota is near exhaustion and missing default resource requests, which can worsen scheduling and autoscaling behavior.

Run Metadata

Identity
Run ID
abd9d69c
Artifact family
Kubernetes bundle
Normalized UTF-8 JSON manifest containing Kubernetes workload, exposure, RBAC, and status evidence.
Source type
Manual upload
upload
Target ID
cluster:aks-prod-eu-1:namespace:payments
Source label
phase9b-e2e
Collection
Collector
signalforge-collectors
Collector version
1.1.0
Recorded at
Mar 27, 02:24 AM
Analysis
Model
gpt-5.4-mini
Analysis time
19.4s
Tokens used
6,961

Environment Context

Target Host
aks-prod-eu-1Kubernetes (aks)
Kernel
namespace:payments
Uptime
unknown