A CLI tool to facilitate Fleet unit deployments.
The following deployment methods are supported:
- Simple Stop-Start: just stop and start all units in one go
- Rolling Stop-Start: Start and stop units in succession by chunking quantity
- Atomic Switchover: Spawn new units, execute atomic-handler script, then destroy the old units. The atomic-handler is responsible for service discovery updates to make the deployment seamless.
Usage: deploy.py [OPTIONS]
Main function
Options:
--fleet-endpoint TEXT Fleet URI / socket
--name TEXT Name of service to deploy [required]
--tag TEXT Tag label. eg Git tag
--method [stopstart|rolling|atomic]
Deployment method
--instances INTEGER Desired number of instances
--unit-file FILENAME Unit template file
--atomic-handler PATH Program to handle atomic operations
--chunking INTEGER Number of containers to act on each pass. Eg
2
--chunking-percent INTEGER Percentage of containers to act on each
pass. Eg 50
--delay INTEGER Startup delay
--help Show this message and exit.
$ ./deploy.py --name docs --method rolling --chunking-percent 50
*** Rolling Deployment Plan ***
==> Details
Found unit: docs-a.service (launched).
Found unit: docs-b.service (launched).
Chunking: 1 units
==> Steps
Step 1: stop docs-a.service
Step 2: start docs-a.service
Step 3: stop docs-b.service
Step 4: start docs-b.service
==> Run
Starting in 5 seconds... 5 4 3 2 1... Starting.
==> Executing
Stopping docs-a.service.....Done.
Starting docs-a.service..................Done.
Stopping docs-b.service.....Done.
Starting docs-b.service.................Done.
Finished.
- Add wait timeouts for Fleet operations
- Exception handling
- Test coverage
Under active development. Consider it Alpha status.