tf_provider/docs/20_discovery/session_report_2026-02-03.md

Session report — 2026-02-03

Summary

• Worked in /home/naeel/terra/universal_rebuild and test_dummy to validate provider behavior and resource creation for multiple services.
• Used dev override to run Terraform against a locally built provider.
• Focused on Kafka creation and state consistency.

Actions performed

• Ran terraform apply in universal_rebuild/test_dummy with dev override and API token.
• Observed long-running Kafka creation, followed by interruption/operation cancellation during apply.
• Verified Terraform state via terraform show (Kafka was not in state after cancel).
• Confirmed Kafka exists in UI while state lacked it (resource created out-of-band of state).
• Collected service ID list for: Flask(89), Redis(91), MongoDB(92), RabbitMQ(93), NodeJS(95), pgAdmin(96), NodeRed(97), Gitea(99), Mariadb(115), Nifi(117), ApacheKafka(116).

Errors and anomalies

• Terraform apply reported cancellation during Kafka creation; operation marked cancelled while service still continued to create.
• Result: nubes_kafka.test_kafka missing from state, requiring import to reconcile.

Current state

• Existing resources in state: dummy, s3bucket, postgres, vm, lucee.
• Kafka exists in UI but is not yet imported into Terraform state.

Next steps (approved by user)

1. Import Kafka into Terraform state using its instance UUID.
2. Add resources one by one for the listed service IDs and run terraform apply after each addition.

Provider logic changes (2026-02-03)

• Enforced adopt/resume only for ready instances: not created/pending/failed now return errors (no adopt).
• After resume, instance status is re-checked; non-ready states cause an error.
• Resource Read now surfaces non-ready instance status as an error instead of silently keeping state.
• Files updated/regenerated: internal/resources_core/crud.go, tools/gen/main.go, internal/resources_gen/*.