Understanding the AI Platform Training service

This page explains the state of a training cluster through the lifecycle of a training job, and how AI Platform Training handles training errors. You can use this information to adapt your training code accordingly.

Lifecycle of a training job

This section explains how AI Platform Training handles worker VMs through the lifecycle of a training job.

Start workers in parallel

When a training job starts, AI Platform Training schedules as many workers as possible in a short amount of time. As a result, workers may start up in parallel instead of sequentially. In order to reduce startup latency, AI Platform Training starts running your code on each worker as soon as it becomes available. When all the workers are available, AI Platform Training sets the job state to RUNNING.

In most cases, your machine learning framework automatically handles the workers starting in parallel. If you're using a distribution strategy in your training code, you may need to adjust it manually to handle workers starting in parallel. Learn more about distribution strategies in TensorFlow and in PyTorch.

Restart workers during the training job

During a training job, AI Platform Training can restart your master, workers or parameter servers with the same hostname. This can occur for the following reasons:

  • VM maintenance: When the VM running a worker is subjected to VM maintenance, AI Platform Training restarts the worker on another VM. Learn more about live migration for VM maintenance.
  • Non-zero exits: If any worker exits with a non-zero exit code, AI Platform Training restarts that worker immediately in the same VM.

    • If a worker fails due to a common error, it is treated as a permanent error, and AI Platform Training shuts down the entire job. If any containers restart before AI Platform Training shuts down the entire job, these containers may produce logs in Cloud Logging.
    • If a worker fails due to a non-permanent error (any error not listed in the common errors), AI Platform Training allows the restarted worker to continue running, with up to five restarts per worker. After five restarts, if a worker fails again, AI Platform Training retries the entire job up to three times before failing the entire job.

To handle worker restarts in your training code, save checkpoints regularly during training so that you can restore from checkpoints when a worker restarts. If you expect training to take more than four hours, we recommend that you save a checkpoint at least once every four hours. Learn how to use training checkpoints in TensorFlow and in PyTorch.

Successfully completing a job

A training job completes successfully when its primary replica exits with exit code 0. At that point, AI Platform Training shuts down all the other running workers.

How AI Platform Training handles training job errors

This section explains how AI Platform Training handles common training job errors and internal errors.

About one minute after a job ends, AI Platform Training sets the error code on the training job object, based on the exit code.

Handle common errors

AI Platform Training shuts down all workers if it encounters any of the following issues:

Error Type Error Message/Log Note
User code exception The replica REPLICA_NAME exited with a non-zero status of EXIT_CODE. Termination reason: REASON. If the job encountered exit codes that could be transient, AI Platform Training tries to restart the job up to three times. The potentially transient error codes that prompt AI Platform Training to retry the job include the following:
  • SIGABRT
    • ExitCode 6
    • ExitCode 134 (custom containers)
  • SIGSEGV
    • ExitCode 11
    • ExitCode 139 (custom containers)
Out-of-memory The replica REPLICA_NAME ran out of memory and exited with a non-zero status of EXIT_CODE. GKE reserves memory on AI Platform Training nodes. On the smallest machine types (such as n1-standard-4), AI Platform Training system agents can take up to 40% of total memory. For larger VMs, the overhead is relatively small. Compare allocatable memory for n1-standard machine types.
Insufficient capacity in your region (Compute Engine stockout) Resources are insufficient in region: REGION_NAME. Try a different region or use a different accelerator. A stockout happens when Compute Engine is at capacity for your selected CPU or GPU in your region. It is unrelated to your project quota. When this happens, AI Platform Training attempts to restart the job up to three times.

Handle internal errors

If AI Platform Training has an internal error, it attempts to restart a job twice (three attempts in total). If the restart attempts also fail, AI Platform Training returns an internal error with the message: Internal error occurred for the current attempt.