All documentation
Server operations

Commands & recipes

Run a one-off shell command on a server, or save one and run it across several.

One-off commands

Open Command in a server's toolbar, enter a command, and choose whether it runs as the deploy user or as root. It runs in bash, and its output streams to a job like any other operation. The panel keeps the server's full command history, with each command's output and exit status.

Re-run a failed command unchanged from its job in the console. Each run is recorded separately, so the history shows every attempt.

Running as root needs the admin or owner role and a signed-in session. An access token cannot do it, whatever its scopes, and Provisore asks for your password again first.

Recipes

A recipe is a shell script saved on the team, set to run as the deploy user or as root. Write them on the Recipes page, which you open from ⌘K. Run recipe in a server's toolbar runs one on that server. From the Recipes page you can run one on several servers at once. Provisore queues a command per server, each with its own log and outcome, so one list shows which servers succeeded.

A recipe
#!/usr/bin/env bash
set -euo pipefail

apt-get update
apt-get install -y --no-install-recommends redis-tools
redis-cli --version

Every target must be an active server in the team. The commands a recipe creates stay linked to it, including when you re-run one failed target. A root recipe follows the same rules as a root command.

Commands run arbitrary code

A command can do anything on the server, so the audit log records who ran it. Where a managed feature exists, such as a daemon, a cron entry, or a firewall rule, use it instead. Provisore tracks those, and you can change or remove them later from the app.