High Latency and Hallucination After Context Switch (Microservices Migration)

What version of Codex is running?

codex-cli 0.55.0

What subscription do you have?

>_ OpenAI Codex (v0.55.0) │ │ │ │ Visit https://chatgpt.com/codex/settings/usage for up-to-date │ │ information on rate limits and credits │ │ │ │ Model: gpt-5-codex (reasoning high, summaries auto) │ │ Directory: ~/lost-persons-monitor │ │ Approval: on-request │ │ Sandbox: workspace-write │ │ Agents.md: AGENTS.md │ │ Account: smartlearn.developer@gmail.com (Plus) │ │ Session: 019aadf4-18ca-7b93-b4d7-760415989c1d │ │ │ │ Context window: 84% left (54.4K used / 272K) │ │ 5h limit: [██░░░░░░░░░░░░░░░░░░] 11% used (resets 00:10 on 23 Nov) │ │ Weekly limit: [█░░░░░░░░░░░░░░░░░░░] 3% used (resets 19:10 on 29 Nov) │ │ Warning: limits may be stale - start new turn to refresh.

Which model were you using?

Model: gpt-5-codex (reasoning high, summaries auto)

What platform is your computer?

Linux 6.14.0-36-generic x86_64 x86_64

What issue are you seeing?

Codex experienced severe latency and failed to analyze a simple Docker log error (curl: (7) Failed to connect...). After the user initiated a major context change (migrating the entire project to a Docker microservices architecture), Codex seems to have lost the current project context, leading to a non-response/hallucination state.

The specific error Codex was asked to analyze was a network error within the new Docker stack (Failed to connect to connect port 8083), which should have been easily traceable using the previously provided docker-compose.yml and port map.

What steps can reproduce the bug?

Uploaded thread: 019aadf4-18ca-7b93-b4d7-760415989c1d

User provided the full project configuration, including the new docker-compose.yml and a new host port map (40110:3306, 40125:8083, etc.) following a successful microservices migration.

The user then provided the log output for the new connector initialization service (docker compose logs connector_init).

The user's query to Codex was to analyze the error: curl: (7) Failed to connect to connect port 8083 after 0 ms: Couldn't connect to server.

Expected result: Quick diagnosis of the internal Docker network or service readiness issue, based on the known configuration.

Actual result: Severe latency (taking too long to process) followed by a non-contextual, unhelpful response, leading to the "Conversation interrupted" status.

What is the expected behavior?

smartdeveloper@smartdeveloper-iMac12-1:~/lost-persons-monitor$ docker compose logs connector_init
WARN[0000] /home/smartdeveloper/lost-persons-monitor/docker-compose.yml: the attribute version is obsolete, it will be ignored, please remove it to avoid potential confusion
connector_init | Esperando a Kafka Connect para registrar el conector...
connector_init | curl: (7) Failed to connect to connect port 8083 after 0 ms: Couldn't connect to server

Additional information

Tool/Model: Codex (or related OpenAI model)

Thread ID: 019aadf4-18ca-7b93-b4d7-760415989c1d

Context Type: Complex Docker Compose stack with CDC (Debezium/Kafka) and new custom FastAPI microservices.