A service runs an image or a repository. Each deployment starts a new version beside the running one and takes over once it is healthy; variables, scaling, previews, rollbacks and logs are here too.
Start a deployment from a repository's push webhook.
The URL is the credential: it carries a secret only members can read. A push to another branch than the service's deploys a preview if the service has previews on, and nothing otherwise.
Path parameters
iduuidpathrequired
secretstringpathrequired
Responses
202no body
A deployment was queued, or one already was
204no body
The push needed no deployment
404Problem
curl -X POST "https://api.infraflow.app/v1/hooks/deploy/0197b2a4-5c1e-7d3a-9f10-4b6c8e2a1d55/secret" \
-b cookies.txt
Promote a preview: run what a branch's live preview runs at the service's own address, without building again. The service keeps its own variables; the preview stays where it is.
Run a command inside a service's container and wait for it.
The command runs in the first copy of the live deployment, as the user its image runs as. It is a one-off: nothing it starts outlives it once the time is up.
Make new versions start with a share of the requests, or with all of them. With a share, a deployment that is ready answers that part of the traffic beside the live one, until it is given everything or taken back.
Start a stopped service again: what it ran when it was stopped runs again, without building anything, where its image still is. Otherwise it is deployed anew.
Stop a service: its containers are removed and its address stops answering, until it is started again. Its volume, its variables and its history stay. A deployment on its way is given up as well.
Read a service's variables with their values. From the dashboard this needs a recent confirmation of identity (/v1/auth/confirm); without one the answer is 403 confirmation_required.