Remediation Catalogue
Every incident type DataAgent detects, with severity and what the AI analyses.
DataAgent detects issues across all major Kubernetes resource types. This catalogue lists every incident category, its severity, and what context the AI uses when enriching the incident.
Pod incidents
| Incident | Severity | What the AI analyses |
|---|---|---|
| CrashLoopBackOff | Critical | Exit codes, restart history, OOM events, recent log tail |
| OOMKilled | Critical | Memory limits vs. usage trend, pod spec, node memory pressure |
| Image pull failure | Warning | Image name, pull secret validity, registry reachability |
| Pod stuck in Pending | Warning | Node capacity, scheduling constraints, PVC binding status, events |
| Init container failure | Warning | Init container logs, dependency availability |
| High restart count | Warning | Restart frequency, last known error reason |
Deployment & ReplicaSet incidents
| Incident | Severity | What the AI analyses |
|---|---|---|
| Unavailable replicas | Critical | Desired vs. ready count, rollout status, pod-level events |
| Rollout stuck | Warning | Pod template diff, image version, probe configuration |
| Low availability percentage | Warning | Replica ratio, disruption budget, node distribution |
StatefulSet incidents
| Incident | Severity | What the AI analyses |
|---|---|---|
| Pod not ready | Critical | Readiness probe configuration, PVC status, volume mounts |
| PVC binding failure | Critical | StorageClass availability, capacity, access mode compatibility |
DaemonSet incidents
| Incident | Severity | What the AI analyses |
|---|---|---|
| Desired vs. ready mismatch | Warning | Node selector, tolerations, pod scheduling failures |
Node incidents
| Incident | Severity | What the AI analyses |
|---|---|---|
| MemoryPressure | Critical | Node memory capacity vs. allocatable, top memory-consuming pods |
| DiskPressure | Critical | Node disk usage, log volume growth, PVC usage |
| PIDPressure | Warning | Process count, workloads with high process counts |
| Node NotReady | Critical | Kubelet status, condition history, recent node events |
Job & CronJob incidents
| Incident | Severity | What the AI analyses |
|---|---|---|
| Job failed | Warning | Exit code, backoff limit, pod error reason |
| Missed scheduled run | Warning | Last schedule time, concurrency policy, pod logs from last run |
PersistentVolumeClaim incidents
| Incident | Severity | What the AI analyses |
|---|---|---|
| PVC unbound | Critical | StorageClass, requested access mode, available PV capacity |
What the AI cannot detect
DataAgent analyses Kubernetes-observable conditions. It does not detect:
- Application-level errors not reflected in pod status or events
- Conditions inside custom resources not covered by K8sGPT analyzers
- Network policies blocking traffic (no packet inspection)
- Security vulnerabilities in container images