Skip to content

Troubleshooting

Consolidated fixes across all clouds. Still stuck? Email contact@argmin.co with the failing command's output — onboarding is white-glove.

AWS

Symptom Most likely cause
Argmin reports "AccessDenied" assuming the role external_id mismatch — confirm with your onboarding contact.
Argmin reports CUR data is missing CUR bucket name or prefix doesn't match terraform.tfvars.
Argmin reports Bedrock data is missing enable_bedrock_access = false, or Bedrock invocation logging isn't enabled. See AWS docs.
terraform apply fails on iam:CreateRole Caller lacks IAM admin. Run as a user with IAMFullAccess or equivalent.

Google Cloud

Symptom Most likely cause
"unauthorized_client" / "subject not allowed" wif_allowed_subjects is missing the exact subject Argmin signs with. Exact-string, no wildcards.
BigQuery query returns empty billing_dataset_id is set but the dataset doesn't exist or isn't the billing export. Check bq ls.
Vertex AI logs missing Cloud Logging exclusion filters are dropping Vertex AI rows before the sink Argmin reads.
terraform apply fails creating the WIF pool Caller lacks roles/iam.workloadIdentityPoolAdmin. Have an admin grant it, or apply from a CI runner that has it.

Azure

Symptom Most likely cause
"AADSTS70021" / "no matching federated identity record" federated_credential_subject mismatch — Argmin's signing subject doesn't match the module.
Cost Management returns empty Tenant-scoped permissions for EA/CSP customers; ask Argmin whether your enrollment account ID must be added separately.
Azure OpenAI usage missing Inference is going through APIM or an Event Hub; flip the matching enable_* flag and terraform apply.
terraform apply fails on azuread_application Terraform principal lacks Application.ReadWrite.OwnedBy. Run as a Global Administrator or have an admin grant it.

verify.sh exits non-zero

What happened Meaning Do this
A read probe failed (403) Permission gap — a granted role didn't actually apply Re-check terraform apply succeeded; share the output with Argmin.
A write probe succeeded Security finding — the identity has more than read access Stop. Do not hand outputs to Argmin; contact your onboarding rep immediately.

Ingestion API

Symptom Cause
401 Unauthorized Missing/invalid Authorization: Bearer or X-API-Key.
422 Unprocessable Entity Payload failed schema validation. Check the required fields (event_id, ingestion_layer, ingestion_timestamp, provider, requested_model) and that estimated_cost_usd is a string.
Event accepted (202) but not in dashboard Allow ~30s; confirm you're querying the same tenant/instance.

Proxy

Symptom Cause
Calls work but aren't attributed The proxy records the call but has no identity hint — add the identity header your platform team configured.
Latency concern The proxy is fail-open on a hard budget; if it's ever slow it forwards anyway. Report sustained latency to Argmin.

Contact support