59f96651b6
For #26713 # Details This PR updates Fleet and its related tools and binaries to use Go version 1.24.1. Scanning through the changelog, I didn't see anything relevant to Fleet that requires action. The only possible breaking change I spotted was: > As [announced](https://tip.golang.org/doc/go1.23#linux) in the Go 1.23 release notes, Go 1.24 requires Linux kernel version 3.2 or later. Linux kernel 3.2 was released in January of 2012, so I think we can commit to dropping support for earlier kernel versions. The new [tools directive](https://tip.golang.org/doc/go1.24#tools) is interesting as it means we can move away from using `tools.go` files, but it's not a required update. # Checklist for submitter If some of the following don't apply, delete the relevant line. <!-- Note that API documentation changes are now addressed by the product design team. --> - [X] Changes file added for user-visible changes in `changes/`, `orbit/changes/` or `ee/fleetd-chrome/changes`. - [x] Manual QA for all new/changed functionality - For Orbit and Fleet Desktop changes: - [X] Make sure fleetd is compatible with the latest released version of Fleet - [x] Orbit runs on macOS ✅ , Linux ✅ and Windows. - [x] Manual QA must be performed in the three main OSs, macOS ✅, Windows and Linux ✅. |
||
|---|---|---|
| .. | ||
| fleetdm_client | ||
| provider | ||
| tf | ||
| .gitignore | ||
| generator.yaml | ||
| go.mod | ||
| go.sum | ||
| main.go | ||
| Makefile | ||
| openapi.json | ||
| README.md | ||
Terraform Provider for FleetDM Teams
This is a Terraform provider for managing FleetDM teams. When you have 100+ teams in FleetDM, and manually managing them is not feasible. The primary setting of concern is the team's "agent options" which consists of some settings and command line flags. These (potentially dangerously) configure FleetDM all machines.
Usage
All the interesting commands are in the Makefile. If you just want
to use the thing, see make install and make apply.
Note that if you run terraform apply in the tf directory, it won't
work out of the box. That's because you need to set the
TF_CLI_CONFIG_FILE environment variable to point to a file that
enables local development of this provider. The Makefile does this
for you.
Future work: actually publish this provider.
Development
Code Generation
See make gen. It will create team_resource_gen.go, which defines
the types that Terraform knows about. This is automatically run
when you run make install.
Running locally
See make plan and make apply.
Running Tests
You probably guessed this. See make test. Note that these tests
require a FleetDM server to be running. The tests will create teams
and delete them when they're done. The tests also require a valid
FleetDM API token to be in the FLEETDM_APIKEY environment variable.
Debugging locally
The basic idea is that you want to run the provider in a debugger. When terraform normally runs, it will execute the provider a few times in the course of operations. What you want to do instead is to run the provider in debug mode and tell terraform to contact it.
To do this, you need to start the provider with the -debug flag
inside a debugger. You'll also need to give it the FLEETDM_APIKEY
environment variable. The provider will print out a big environment
variable that you can copy and paste to your command line.
When you run terraform apply or the like, you'll invoke it with
that big environment variable. It'll look something like
TF_REATTACH_PROVIDERS='{"fleetdm.com/tf/fleetdm":{"Protocol":"grpc","ProtocolVersion":6,"Pid":33644,"Test":true,"Addr":{"Network":"unix","String":"/var/folders/32/xw2p1jtd4w10hpnsyrb_4nmm0000gq/T/plugin771405263"}}}' terraform apply
With this magic, terraform will look to your provider that's running in a debugger. You get breakpoints and the goodness of a debugger.