πŸ”§CLI Node configuration

Once Coretex CLI Tool is installed, you need to run the initial configuration for the user and the node separately. You will be prompted with several steps required to authenticate and configure your Coretex environment.

Configuring Coretex CLI user allows you to manually query Coretex API server to perform various actions directly from your terminal, such as create datasets, list the runs, or download models.

If you only wish to connect your local computer as an execution Node to Coretex servers without using CLI for querying the Coretex API manually from your terminal, you do not have to configure CLI user, only the node.

User configuration

To start the user configuration process, enter this command in the terminal.

coretex config --user

Enter your account credentials

These inputs will be prompted one after another. That means that you will need to enter your Coretex account email first and then your password.

Storage path

Path that will be used for storing downloaded datasets and samples. Press Enter to use default path (recommended) or optionally enter different storage path.

Node configuration

To start the node configuration process, enter this command in the terminal:

coretex config --node

Node name

The first step after entering the previous command is to enter your node name. This will be displayed in the web UI under the Nodes section.

Storage path

Path that will be used for storing downloaded datasets and samples locally on the node. Press enter to use default path (recommended) or optionally enter different storage path.

Note: Storage path should be the same as (and if) used during --user config

Choose GPU or CPU image

To be able to select a Docker image to use with your Coretex Node you will need to use up and down arrows on your keyboard.

RAM memory limit

Choose the amount of RAM memory Coretex Node container will be allowed to use. This can't be higher that your system RAM.

Node swap memory limit

Enter the Node swap memory limit in GB. Swap is the storage (hard drive) space a process will use when it runs out of RAM in order to try to keep the processes running.

A good recommendation is to make this larger than RAM memory limit.

Please note that run which uses swap memory is very likely to run for unusually long time and most probably fail or timeout. It is crucial to monitor the metrics during the run execution, so you can either tune the run parameters to lower the amount of RAM it uses or execute the run on the Node with more system RAM to avoid using swap.

Node shared memory (SHM) limit

Enter Node SHM limit in GB or press enter to use default. Docker uses shared memory to provide a temporary file storage filesystem using RAM for storing files. It also facilitates inter-process communication (IPC). The shared memory device, /dev/shm, is optional but it’s probably desirable since it facilitates IPC.

If your runs are crashing with shared memory errors, try increasing this limit to accommodate shared memory requirements.

Last updated