Version v0.5.6 of the documentation is no longer actively maintained. The site that you are currently viewing is an archived snapshot. For up-to-date documentation, see the latest version.

dexaflow sync

Make the target run the DAGs of a workspace: deploy what changed, report or pause what is gone (Pro).

Synopsis

Compares the DAG projects under

with the DAGs registered on the target in use and makes the target match:

in sync: the local content label is the current version; nothing is done.
drift or not deployed: the content label is not registered; the DAG is deployed exactly as dexaflow deploy --all deploys it (build or reuse the image, push, register). A new DAG is paused or not as the server decides for new DAGs; a redeploy never changes is_paused.
behind: the content label is a registered version older than the current one, so the target runs something newer than this checkout (it was deployed from a newer commit). It is reported and left alone; with –rollback that older version is made current again, as dexaflow rollback --to <label> does, with nothing built and is_paused kept.
not in workspace: the target holds a DAG no project underdefines. It is reported, and with –prune it is paused (reversible with dexaflow admin dags unpause). Nothing is ever deleted. Every DAG the token can see on the target is a candidate: sync one folder per target.

–prune applies nothing while a project cannot be read (a dexaflow.yaml that does not parse hides its dag_id), or when it would pause more DAGs than –max-prune; without a terminal it needs –yes.

Targets, tokens and deploy defaults resolve as for a command run in

: –server, –target, then the target selected for the workspace around, its default_target, then the config file. The labels are computed as deploy computes them (git describe, the content and the CLI version), so pass the –registry, –platform and –dockerfile a deploy is given.

–dry-run builds, pushes, registers and changes nothing, so it can run as a CI check or an Argo CD hook.

Exit codes: 0 when the target matches the workspace (–dry-run) or every change was applied; 1 when –dry-run found something to change, or a behind DAG was left alone without –rollback; 2 when the target or a project could not be read, a change failed or was refused, the confirmation was declined, or the command was used wrongly. A DAG not in the workspace counts as a change only with –prune, and only while it is not paused.

dexaflow sync <dir> [flags]

Options

      --builder string      image build tool to shell out to (e.g. docker, podman, nerdctl) (default "docker")
      --dockerfile string   Dockerfile path relative to the DAG directory (default "Dockerfile")
      --dry-run             say per DAG what sync would do and change nothing; exit 1 when something would change
  -h, --help                help for sync
      --max-prune int       with --prune, apply nothing when more than this many DAGs would be paused (default: the number of DAG projects under <dir>) (default -1)
      --output string       output format: text, or json for one JSON document on stdout (human lines go to stderr) (default "text")
      --platform strings    image platform to build for, os/arch (repeatable or comma-separated), over build.platforms in dexaflow.yaml, the target and the workspace defaults (default linux/amd64)
      --prune               pause the DAGs the target holds and the workspace does not (never deletes)
      --registry string     image registry prefix to push to, over every other source (then DEXAFLOW_REGISTRY, dexaflow.yaml, the target, the workspace defaults, the config file, the cluster's deploy defaults)
      --rollback            make current again the older registered version a behind DAG's content equals (default: report it and leave it)
      --server string       control plane base URL, over --target (default: the target in use, then server_url from the config file)
      --token string        JWT bearer token (default: DEXAFLOW_TOKEN, then the token saved for the server)
  -y, --yes                 skip the confirmation prompt (for automation; required for --prune without a terminal)

Options inherited from parent commands

      --config string       config file path (default ~/.dexaflow/config.yaml)
      --log-level string    log level: debug, info, warn, error
      --server-url string   control plane API base URL
      --target name         name of the control plane to use (default: DEXAFLOW_TARGET, then the target selected for the workspace; see dexaflow target list)

dexaflow sync

Synopsis

Options

Options inherited from parent commands

SEE ALSO