
Here, you will find information on running the Cedana daemon on your machine, system architecture, and the various features of both the daemon and CLI.
The daemon is designed to manage the lifecycle of processes/containers, including checkpoint/restore, in the larger Cedana system. However, it can be installed and used independently as a C/R tool with its convenient defaults and a friendly command-line interface.
For detailed documentation on our managed Kubernetes or the larger Cedana system, please see here.
Quick start
First, ensure that you have Cedana installed on your machine, and the daemon is running. See installation.
Run a new managed process:
cedana run process --attach test/workloads/date-loop.sh
Any process/container you spawn using cedana run
creates a managed job. To view all managed jobs:
cedana ps
JOB TYPE PID STATUS GPU CHECKPOINT SIZE LOG
personal_hopper9 process 32646 sleep no [Attachable]
Checkpoint the job:
cedana dump job personal_hopper9
If you view the jobs again, you will see that it was checkpointed:
JOB TYPE PID STATUS GPU CHECKPOINT SIZE LOG
personal_hopper9 process 32646 halted no 2 seconds ago 644 KiB
Restore the job:
cedana restore job --attach personal_hopper9
For all available CLI options, see CLI reference. Directly interacting with daemon is also possible through gRPC, see API reference.
For specific usage, check out the guides. For information on architecture or to get started with contributing, check out the developer guides.
Get started
Guides
Developer guides
References
Last updated
Was this helpful?