| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| |
|
|
|
|
|
|
|
|
| |
- Add ValidateName() enforcing ^[a-z][a-z0-9-]{0,62}$ on all entry points
- Tighten sudoers to restrict cp sources/destinations and chown targets
- Scope git safe.directory to www-data user only (preserves CVE-2022-24765)
- Add www-data to git group and caddy to www-data group for fcgiwrap
- Fix vanity import template to use orig_uri placeholder
- Restart (not reload) services after group changes
- Add name validation to env subcommands and deploy_cmd
|
| |
|
|
|
| |
The pipe caused only echo to run as sudo while tee had no
privileges. Wrap in sh -c so the redirect runs as root.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
New deployment model where projects start with a git remote on the VPS.
Pushing to the remote triggers automatic docker build and deploy via
post-receive hooks. The base domain serves Go vanity imports and git
HTTPS cloning via Caddy + fcgiwrap.
- Add `ship init <name>` command to create bare repos and .ship/ config
- Add `ship deploy <name>` command for manual rebuilds
- Extend `ship host init --base-domain` to set up Docker, git user,
fcgiwrap, sudoers, and vanity import infrastructure
- Add git-app and git-static types alongside existing app and static
- Update remove, status, logs, restart, list, and config-update to
handle new types
|
| |
|
|
|
|
|
| |
Restructure CLI files to follow idiomatic cobra layout:
- main.go: minimal entry point
- root.go: command definition, flags, and subcommand wiring
- deploy.go: all deploy implementation
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Previously, updating an app or static site without specifying all flags
would clear existing configuration (e.g., custom domains would be lost
if --domain wasn't provided on update).
Now all config is merged in runDeploy before calling deploy functions:
- Load state once instead of 2-3 times
- Check for existing app once to determine create vs update
- Build merged DeployOptions with existing config + CLI overrides
- Deploy functions receive fully-merged config and just execute
This ensures existing configuration is preserved unless explicitly
overridden by CLI flags.
|
| | |
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
| |
Adds --memory and --cpu flags to set systemd resource limits:
ship --binary ./app --memory 512M --cpu 100%
Also adds config update mode - use --name without --binary to
update an existing app's config without redeploying the binary:
ship --name myapp --cpu 50%
ship --name myapp --memory 256M --env DEBUG=true
Limits are stored in state and preserved on redeploy.
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
When a base domain is configured on a host (e.g., apps.example.com),
deployments automatically get a subdomain ({name}.apps.example.com).
Custom --domain can still be provided to route both domains.
- Add BaseDomain field to Host state
- Add --base-domain flag to host init
- Add 'ship host set-domain' command to update base domain
- Update deploy flow to auto-generate subdomains
- Fix error display (errors were being silently swallowed)
- Remove placeholder email from Caddyfile template
|
| |
|
|
|
| |
Show a ship logo in CLI help and explain how ship works -
SSH-only deployment with no agents or containers required.
|
|
|
- Rename module to github.com/bdw/ship
- Rename cmd/deploy to cmd/ship
- Update all import paths
- Update config path from ~/.config/deploy to ~/.config/ship
- Update VPS env path from /etc/deploy to /etc/ship
- Update README, Makefile, and docs
|
