Common Errors Using runai-bgu CLI
Authentication Issues
If you encounter one of the following errors:
$ runai-bgu submit ... Run:AI token has expired, please re-login using: runai-bgu login
OR
$ runai-bgu submit ...
Get "https://172.20.7.230:6443/api/v1/namespaces?labelSelector=runai%2Fqueue%3Dusername": failed to refresh token: oauth2: "invalid_grant" "Token is not active”Solution:
To resolve this issue, you need to re-authenticate with the Run:AI platform by executing the runai-bgu login command.
Follow the on-screen instructions to complete the authentication process.
For detailed guidance on using the runai-bgu login command, refer to the Configure Run:AI CLI Tool guide.
Job Submission Issues:
Job Name
$ runai-bgu submit ... Error: Error: invalid workload name. string contain other characters other then lower case alphanumeric or hyphen, but not running. Please resume or delete job and try again.
Solution:
This error indicates that the job name contains invalid characters. Ensure that the job name consists only of lowercase alphanumeric characters or hyphens. Run the command again with a valid job name.
CPU Limit
$ runai-bgu submit ... CPU limit must be set, CPU limit must be a positive integer.
Solution:
This error indicates that the CPU limit is not set. Ensure that you specify a positive integer for the CPU limit when submitting the job.
$ runai-bgu submit ... Invalid input for CPU limit, must be a positive integer.
Solution:
This error indicates that the CPU limit provided is not a valid positive integer. Ensure that you enter a valid positive integer for the CPU limit when submitting the job.
Memory Limit
$ runai-bgu submit ... Memory limit must be set, memory limit cannot be empty. Memory must end in a unit [K, M, G, Ki, Mi, Gi].
Solution:
This error indicates that the memory limit is not set or is incorrectly formatted. Ensure that you specify a valid memory limit that ends with a unit (K, M, G, Ki, Mi, Gi) when submitting the job.
$ runai-bgu submit ... Invalid input for memory limit, memory must end in a unit [K, M, G, Ki, Mi, Gi]
Solution:
This error indicates that the memory limit provided does not end with a valid unit. Ensure that you enter a valid memory limit that ends with one of the specified units (K, M, G, Ki, Mi, Gi) when submitting the job.
GPU Limit
$ runai-bgu submit ... <Job Name> is a CPU only job and should not be started with a GPU
Solution:
This error indicates that the job is configured as CPU-only and cannot be submitted with a GPU limit. Ensure that you do not specify a GPU limit for CPU-only jobs. If you intend to run a GPU job, ensure you are using the correct job type to include GPU resources.
Workload Name
$ runai-bgu submit ... Error: failed to submit. Workload name <workload-type> already exists in the project, but not running. Please resume or delete job and try again.
Solution:
This error indicates that a workload with the specified name already exists in the project but is not currently running. To resolve this issue, you can either resume the existing job or delete it and then try submitting the new job again. Use the runai-bgu resume <job-name> command to resume the existing job, or the runai-bgu delete <job-name> command to delete it.