Skip to main content

Troubleshooting

Triage order

Work from the outside in. Start with a wide view, then drill into the failing resource.

# 1. Wide view: operator, all platforms, managed infra, connectors, proxies.
$ ilmctl status -v

# 2. Analyzer findings with remediation hints.
$ ilmctl check

# 3. Drill into a specific Platform.
$ ilmctl platform describe ilm

# 4. Tail component logs.
$ ilmctl platform logs --component core -f

# 5. List events for the Platform and its children.
$ ilmctl platform events ilm

check runs the same analyzer engine as diagnostics analyze, so a live diagnosis and an offline bundle analysis produce identical findings. It exits non-zero when any finding has fail severity.

Common failures

SymptomLikely causeSuggested remediation
DatabaseReady=FalseManaged CloudNativePG Cluster unhealthyInspect the CNPG Cluster named in the finding; check storage quotas and node resources
Managed mode, CRD not foundUpstream operator absentilmctl deps install --only cnpg (or the appropriate dep)
secretRef / issuerRef to absent objectMisnamed Secret or Issuer reference in the Platform specCreate the Secret or Issuer, or correct the reference in the Platform spec
Platform stuck in ProgressingReconcile blocked or operator unhealthyilmctl platform events ilm, then ilmctl platform logs --component core
observedGeneration < generationOperator has not yet observed the latest spec changeVerify the operator Deployment is Running and has no restart loops
Connector waitingForApprovalPlatform not yet reachableEnsure the Platform is Running and the connector's --platform-url is correct

Component log targets

ilmctl platform logs accepts a --component flag. Available components:

core, auth, auth-opa-policies, scheduler, fe-administrator, utils, api-gateway, provisioning-rabbitmq (only when provisioning.mode=deploy).

Getting a machine-readable diagnosis

$ ilmctl check -o json | jq '.[] | select(.severity == "fail")'

For a full offline bundle to hand to support, see Diagnostics.