First pipleline fails | CPU quota exceeded - but it doesn't appear to be?

[mary99]

Hello there,

I have signed up for the free trial of Google Cloud with $300 credit. I wanted to create my first AutoML model but cannot get passed the pipeline as it keeps failing. I have uploaded my CSV file to create a new dataset, from there I clicked "Create Model". In the model settings I made sure to set choose a machine type that is within my quota (and to only use one type), and I set the maximum number of CPUs to 2 (my quota maximum appears to be 8). Below is a screenshot showing all details as well as the error. 

Does anyone perhaps know where I have gone wrong?

com.google.cloud.ai.platform.common.errors.AiPlatformException: code=RESOURCE_EXHAUSTED, message=The following quota metrics exceed quota limits: aiplatform.googleapis.com/custom_model_training_cpus, cause=null; Failed to create custom job for the task. Task: Project number: 977210408547, Job id: 1938217173298184192, Task id: -4929050085460803584, Task name: set-optional-inputs, Task state: DRIVER_SUCCEEDED, Execution name: projects/977210408547/locations/us-central1/metadataStores/default/executions/7171705058993674372; Failed to create external task or refresh its state. Task:Project number: 977210408547, Job id: 1938217173298184192, Task id: -4929050085460803584, Task name: set-optional-inputs, Task state: DRIVER_SUCCEEDED, Execution name: projects/977210408547/locations/us-central1/metadataStores/default/executions/7171705058993674372; Failed to handle the pipeline task. Task: Project number: 977210408547, Job id: 1938217173298184192, Task id: -4929050085460803584, Task name: set-optional-inputs, Task state: DRIVER_SUCCEEDED, Execution name: projects/977210408547/locations/us-central1/metadataStores/default/executions/7171705058993674372

[nceniza]

I believe it's about to do with the regions resource availability. I would suggest trying to reduce the nodes for your resources (Probably choosing n2-standard-2 machine type instead), another one is try a different region when running the pipeline and lastly retry on a different time as it seems these trial resources are already exhausted at the time of execution.