> See the full index of documentation pages at https://docs.zettabyte.com/llms.txt.

# Troubleshooting

> Find a fix by symptom: pages titled by the error or symptom, a list of common issues, and one reference page for the supported framework versions.

Source: https://docs.zettabyte.com/docs/zcloud/troubleshooting

Last updated: 2026-11-05

## Pages

- [Common issues](https://docs.zettabyte.com/docs/zcloud/troubleshooting/common-issues.md): Find your symptom, then apply the fix. Grouped by resources, storage, serving, training, images, and environment.
- [Cannot mount two volumes on one task](https://docs.zettabyte.com/docs/zcloud/troubleshooting/cannot-mount-two-volumes.md): When a zCLOUD task cannot mount two volumes, check the mount directories: mount directories must be unique within a task.
- [Distributed task fails immediately](https://docs.zettabyte.com/docs/zcloud/troubleshooting/distributed-task-fails-immediately.md): When a training task on zCLOUD, the GPU cloud, is set to PyTorch or MPI, check its framework: the framework sets what kind of job the task is.
- [TorchDynamo compilation errors](https://docs.zettabyte.com/docs/zcloud/troubleshooting/torchdynamo-compilation-errors.md): A training job that fails with a TorchDynamo compilation error from `torch.compile` can have TorchDynamo turned off with one environment variable on the task.
- [Environment and versions](https://docs.zettabyte.com/docs/zcloud/troubleshooting/environment-and-versions.md): The library version recorded for the platform environment.

## Optional

- [Full Troubleshooting documentation](https://docs.zettabyte.com/docs/zcloud/troubleshooting/llms-full.txt): every Troubleshooting page as one markdown file
