Olares

English

简体中文

English

简体中文

Appearance

Model or engine is not ready ​

Use this guide when a model instance opens in Model Console but Model does not become Ready, or Engine does not become Running.

Identify which stage is blocked ​

Open the model instance from the Launchpad and check Service status on the Status tab.

What you seeWhat it means
Model is not ReadyThe model files are still downloading or the download and verification process could not finish.
Model is Ready, but Engine is not RunningThe files are available, but the inference service could not start with the current configuration or resources.
The model app is Stopped in Market or SettingsOlares could not reserve the resources required to launch the app, or the app was stopped manually.

If the model is not ready ​

  1. If the model is still downloading, leave the model instance running. Large model files can take time to download.
  2. Confirm that Olares can reach the model source and has enough free disk space. If storage is low, see Free up disk space.
  3. For an instance created from an Engine Base app, compare its MODEL_SOURCE and MODEL_NAME values with the format required by that engine. See Configure engine environment variables.
  4. If you correct an environment variable in Settings > Applications > [model app] > Manage environment variables, save the change and click Apply. Wait for Model Console to report the new status.

Do not delete partially downloaded model files while the instance is running. This can leave the app and its stored files out of sync.

If the engine is not running ​

  1. Wait until Model shows Ready. The engine cannot serve a model whose files are not ready.
  2. Review the instance configuration:
    • Confirm that MODEL_NAME matches the configured model.
    • Check that ENGINE_ARGS use the syntax required by the selected engine.
    • Check that the engine's REQUIRED_GPU_MEMORY value does not exceed the available resource.
  3. If the app is Stopped, resume it from Market > My Olares or Settings > Applications.
  4. If Olares reports insufficient GPU memory, follow GPU app remains stopped after installation or resume.
  5. Return to Model Console. Continue only after Model shows Ready and Engine shows Running.

If the status does not change ​

Do not delete the app namespace, Application resource, or shared engine service. These actions can affect other model instances and make recovery harder.

Record the following information:

  • Olares version and model app version.
  • Engine Base type and whether the app is a pre-built model or a cloned instance.
  • The exact Model and Engine status.
  • The model source and engine arguments, with tokens or credentials removed.
  • The time the failure started and any change made immediately before it.

Then download the application operation log from Market > My Olares > Logs. If more detail is requested, follow Collect diagnostic information and share the archive through a private support channel.