support agent on windows

.github/workflows/build-msi.yaml

@@ -0,0 +1,46 @@
+name: Build MSI
+
+on:
+  workflow_dispatch:
+    inputs:
+      # since this is being triggered manually on branch but we should use our regular git tags when we're back on the normal flow
+      msi_version:
+        description: "MSI package version (e.g., 1.0.0)"
+        required: true
+
+jobs:
+  build-msi:
+    runs-on: windows-2019
+
+    steps:
+      - name: Checkout code
+        uses: actions/checkout@v4
+
+      - name: Set up Go
+        uses: actions/setup-go@v4
+        with:
+          go-version: '1.23.1'
+
+      - name: Install Make
+        run: choco install make --yes
+
+      - name: Install WiX CLI
+        run: dotnet tool install --global wix
+
+      - name: Install WiX Extensions
+        run: |
+          wix extension add -g WixToolset.Firewall.wixext/5.0.2
+          wix extension add -g WixToolset.Util.wixext/5.0.2
+
+      - name: Build Go binary
+        run: make windows
+
+      - name: Build MSI
+        run: |
+          wix build agent.wxs -define GoBinDir="${{ github.workspace }}\bin" -define MSIProductVersion="${{ github.event.inputs.msi_version }}" -ext WixToolset.Util.wixext -ext WixToolset.Firewall.wixext -o agent-${{ github.event.inputs.msi_version }}.msi
+
+      - name: Upload MSI artifact
+        uses: actions/upload-artifact@v4
+        with:
+          name: agent-${{ github.event.inputs.msi_version }}.msi
+          path: agent-${{ github.event.inputs.msi_version }}.msi

Makefile

@@ -16,7 +16,7 @@ else
 PATH_VERSION = v$(TAG_VERSION)
 endif
 
-LDFLAGS = "-s -w -X 'github.com/viamrobotics/agent/utils.Version=${TAG_VERSION}' -X 'github.com/viamrobotics/agent/utils.GitRevision=${GIT_REVISION}'"
+LDFLAGS = "-s -w -X 'github.com/viamrobotics/agent.Version=${TAG_VERSION}' -X 'github.com/viamrobotics/agent.GitRevision=${GIT_REVISION}'"
 TAGS = osusergo,netgo
 
 
@@ -33,16 +33,22 @@ arm64:
 amd64:
 	make GOARCH=amd64
 
-bin/viam-agent-$(PATH_VERSION)-$(LINUX_ARCH): go.* *.go */*.go */*/*.go *.service Makefile
+bin/viam-agent-$(PATH_VERSION)-$(LINUX_ARCH): go.* *.go */*.go */*/*.go subsystems/viamagent/*.service Makefile
 	go build -o $@ -trimpath -tags $(TAGS) -ldflags $(LDFLAGS) ./cmd/viam-agent/main.go
 	test "$(PATH_VERSION)" != "custom" && cp $@ bin/viam-agent-stable-$(LINUX_ARCH) || true
 
+.PHONY: windows
+windows: bin/viam-agent.exe
+
+bin/viam-agent.exe:
+	GOOS=windows GOARCH=amd64 go build -o $@ -trimpath -tags $(TAGS) -ldflags $(LDFLAGS) ./cmd/viam-agent
+
 .PHONY: clean
 clean:
 	rm -rf bin/
 
 bin/golangci-lint: Makefile
-	GOBIN=`pwd`/bin go install github.com/golangci/golangci-lint/cmd/golangci-lint@v1.64.5
+	GOBIN=`pwd`/bin go install github.com/golangci/golangci-lint/cmd/golangci-lint@v1.60.3
 
 .PHONY: lint
 lint: bin/golangci-lint

agent.bat

@@ -0,0 +1,16 @@
+@echo off
+:: installer for agent on windows
+
+set root=\opt\viam
+set fname=viam-agent-windows-amd64-alpha-16-6dece14.exe
+mkdir %root%\cache
+mkdir %root%\bin
+curl https://storage.googleapis.com/packages.viam.com/temp/%fname% -o %root%\cache\%fname%
+netsh advfirewall firewall add rule name="%fname%" dir=in action=allow program="c:\%root%\cache\%fname%" enable=yes
+del %root%\bin\viam-agent.exe
+mklink %root%\bin\viam-agent.exe %root%\cache\%fname%
+:: todo: restart on error
+sc create viam-agent binpath= c:%root%\bin\viam-agent.exe start= auto
+sc failure viam-agent reset= 0 actions= restart/30000/restart/30000/restart/30000
+sc failureflag viam-agent 1
+sc start viam-agent

agent.wxs

@@ -0,0 +1,28 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<Wix xmlns="http://wixtoolset.org/schemas/v4/wxs"
+    xmlns:util="http://wixtoolset.org/schemas/v4/wxs/util"
+    xmlns:fire="http://wixtoolset.org/schemas/v4/wxs/firewall">
+    <Package Name="viam-agent" Manufacturer="viam" Version="$(var.MSIProductVersion)" UpgradeCode="d3b5bca3-4bec-46cb-a063-ca6315de7de4" Language="1033" Scope="perMachine">
+        <Media Id="1" Cabinet="media1.cab" EmbedCab="yes" />
+
+        <StandardDirectory Id="TARGETDIR">
+            <Directory Id="CustomInstallPath" Name="opt">
+                <Directory Id="ViamFolder" Name="viam">
+                    <Directory Id="INSTALLFOLDER" Name="bin">
+                        <Component Id="MainServiceComponent" Guid="d478ceba-537c-4e49-9262-bf24ccfe4909">
+                            <File Id="ViamExe" Source="$(var.GoBinDir)\viam-agent.exe" KeyPath="yes" />
+                            <ServiceInstall Id="InstallAgentervice" Name="viam-agent" DisplayName="viam-agent Service" Description="viam-agent Windows service" Start="auto" Type="ownProcess" ErrorControl="normal" Account="LocalSystem" Interactive="yes" />
+                            <ServiceControl Id="ControlAgentService" Name="viam-agent" Start="install" Stop="both" Remove="uninstall" Wait="yes" />
+                            <fire:FirewallException Id="AllowAllTCP" Name="viam-agent" Profile="all" Protocol="tcp" Scope="any" />
+                            <fire:FirewallException Id="AllowAllUDP" Name="viam-agent" Profile="all" Protocol="tcp" Scope="any" />
+                        </Component>
+                    </Directory>
+                </Directory>
+            </Directory>
+        </StandardDirectory>
+
+        <Feature Id="MainFeature" Title="Main Feature" Level="1">
+            <ComponentRef Id="MainServiceComponent" />
+        </Feature>
+    </Package>
+</Wix>

cmd/viam-agent/main.go

@@ -4,12 +4,11 @@ import (
 	"bytes"
 	"context"
 	"fmt"
-	"io/fs"
 	"os"
-	"os/exec"
 	"os/signal"
 	"os/user"
 	"path/filepath"
+	"runtime"
 	"strings"
 	"sync"
 	"syscall"
@@ -19,12 +18,10 @@ import (
 	"github.com/nightlyone/lockfile"
 	"github.com/pkg/errors"
 	"github.com/viamrobotics/agent"
-	"github.com/viamrobotics/agent/subsystems/networking"
 	_ "github.com/viamrobotics/agent/subsystems/syscfg"
 	"github.com/viamrobotics/agent/utils"
 	"go.uber.org/zap"
 	"go.viam.com/rdk/logging"
-	goutils "go.viam.com/utils"
 )
 
 var (
@@ -34,27 +31,28 @@ var (
 	globalLogger = logging.NewLogger("viam-agent")
 )
 
+//nolint:lll
+type agentOpts struct {
+	Config         string `default:"/etc/viam.json"                        description:"Path to machine credentials file"   long:"config"   short:"c"`
+	DefaultsConfig string `default:"/etc/viam-defaults.json"               description:"Path to manufacturer defaults file" long:"defaults"`
+	Debug          bool   `description:"Enable debug logging (agent only)" env:"VIAM_AGENT_DEBUG"                           long:"debug"    short:"d"`
+	UpdateFirst    bool   `description:"Update versions before starting"   env:"VIAM_AGENT_WAIT_FOR_UPDATE"                 long:"wait"     short:"w"`
+	Help           bool   `description:"Show this help message"            long:"help"                                      short:"h"`
+	Version        bool   `description:"Show version"                      long:"version"                                   short:"v"`
+	Install        bool   `description:"Install systemd service"           long:"install"`
+	DevMode        bool   `description:"Allow non-root and non-service"    env:"VIAM_AGENT_DEVMODE"                         long:"dev-mode"`
+}
+
 //nolint:gocognit
-func main() {
+func commonMain() {
 	ctx, cancel := setupExitSignalHandling()
 
 	defer func() {
 		cancel()
 		activeBackgroundWorkers.Wait()
 	}()
 
-	//nolint:lll
-	var opts struct {
-		Config         string `default:"/etc/viam.json"                        description:"Path to machine credentials file"   long:"config"   short:"c"`
-		DefaultsConfig string `default:"/etc/viam-defaults.json"               description:"Path to manufacturer defaults file" long:"defaults"`
-		Debug          bool   `description:"Enable debug logging (agent only)" env:"VIAM_AGENT_DEBUG"                           long:"debug"    short:"d"`
-		UpdateFirst    bool   `description:"Update versions before starting"   env:"VIAM_AGENT_WAIT_FOR_UPDATE"                 long:"wait"     short:"w"`
-		Help           bool   `description:"Show this help message"            long:"help"                                      short:"h"`
-		Version        bool   `description:"Show version"                      long:"version"                                   short:"v"`
-		Install        bool   `description:"Install systemd service"           long:"install"`
-		DevMode        bool   `description:"Allow non-root and non-service"    env:"VIAM_AGENT_DEVMODE"                         long:"dev-mode"`
-	}
-
+	var opts agentOpts
 	parser := flags.NewParser(&opts, flags.IgnoreUnknown)
 	parser.Usage = "runs as a background service and manages updates and the process lifecycle for viam-server."
 
@@ -87,7 +85,7 @@ func main() {
 	// need to be root to go any further than this
 	curUser, err := user.Current()
 	exitIfError(err)
-	if curUser.Uid != "0" && !opts.DevMode {
+	if runtime.GOOS != "windows" && curUser.Uid != "0" && !opts.DevMode {
 		//nolint:forbidigo
 		fmt.Printf("viam-agent must be run as root (uid 0), but current user is %s (uid %s)\n", curUser.Username, curUser.Uid)
 		return
@@ -98,7 +96,7 @@ func main() {
 		return
 	}
 
-	if !opts.DevMode {
+	if runtime.GOOS != "windows" && !opts.DevMode {
 		// confirm that we're running from a proper install
 		if !strings.HasPrefix(os.Args[0], utils.ViamDirs["viam"]) {
 			//nolint:forbidigo
@@ -141,48 +139,10 @@ func main() {
 	err = manager.LoadAppConfig()
 	//nolint:nestif
 	if err != nil {
-		if cfg.AdvancedSettings.DisableNetworkConfiguration {
-			globalLogger.Errorf("Cannot read %s and network configuration is disabled. Please correct and restart viam-agent.",
-				utils.AppConfigFilePath)
-			manager.CloseAll()
-			return
-		}
-
-		// If the local /etc/viam.json config is corrupted, invalid, or missing (due to a new install), we can get stuck here.
-		// Rename the file (if it exists) and wait to provision a new one.
-		if !errors.Is(err, fs.ErrNotExist) {
-			globalLogger.Error(errors.Wrapf(err, "reading %s", utils.AppConfigFilePath))
-			globalLogger.Warn("renaming %s to %s.old", utils.AppConfigFilePath, utils.AppConfigFilePath)
-			if err := os.Rename(utils.AppConfigFilePath, utils.AppConfigFilePath+".old"); err != nil {
-				// if we can't rename the file, we're up a creek, and it's fatal
-				globalLogger.Error(errors.Wrapf(err, "removing invalid config file %s", utils.AppConfigFilePath))
-				globalLogger.Error("unable to continue with provisioning, exiting")
-				manager.CloseAll()
-				return
-			}
-		}
-
-		// We manually start the provisioning service to allow the user to update it and wait.
-		// The user may be updating it soon, so better to loop quietly than to exit and let systemd keep restarting infinitely.
-		globalLogger.Infof("machine credentials file %s missing or corrupt, entering provisioning mode", utils.AppConfigFilePath)
-
-		if err := manager.StartSubsystem(ctx, networking.SubsysName); err != nil {
-			globalLogger.Error(errors.Wrapf(err, "could not start networking subsystem, "+
-				"please manually update /etc/viam.json and connect to internet"))
+		if !runPlatformProvisioning(ctx, cfg, manager, err) {
 			manager.CloseAll()
 			return
 		}
-
-		for {
-			globalLogger.Warn("waiting for user provisioning")
-			if !goutils.SelectContextOrWait(ctx, time.Second*10) {
-				manager.CloseAll()
-				return
-			}
-			if err := manager.LoadAppConfig(); err == nil {
-				break
-			}
-		}
 	}
 
 	// valid viam.json from this point forward
@@ -200,23 +160,7 @@ func main() {
 		// wait to be online
 		timeoutCtx, cancel := context.WithTimeout(ctx, time.Minute)
 		defer cancel()
-		for {
-			cmd := exec.CommandContext(timeoutCtx, "systemctl", "is-active", "network-online.target")
-			_, err := cmd.CombinedOutput()
-
-			if err == nil {
-				break
-			}
-
-			if e := (&exec.ExitError{}); !errors.As(err, &e) {
-				// if it's not an ExitError, that means it didn't even start, so bail out
-				globalLogger.Error(errors.Wrap(err, "running 'systemctl is-active network-online.target'"))
-				break
-			}
-			if !goutils.SelectContextOrWait(timeoutCtx, time.Second) {
-				break
-			}
-		}
+		waitOnline(globalLogger, timeoutCtx)
 
 		// Check for self-update and restart if needed.
 		needRestart, err := manager.SelfUpdate(ctx)
@@ -269,12 +213,11 @@ func setupExitSignalHandling() (context.Context, func()) {
 			// this will eventually be handled elsewhere as a restart, not exit
 			case syscall.SIGHUP:
 
-			// ignore SIGURG entirely, it's used for real-time scheduling notifications
-			case syscall.SIGURG:
-
 			// log everything else
 			default:
-				globalLogger.Debugw("received unknown signal", "signal", sig)
+				if !ignoredSignal(sig) {
+					globalLogger.Debugw("received unknown signal", "signal", sig)
+				}
 			}
 		}
 	}()
@@ -283,6 +226,7 @@ func setupExitSignalHandling() (context.Context, func()) {
 	return ctx, cancel
 }
 
+// helper to log.Fatal if error is non-nil.
 func exitIfError(err error) {
 	if err != nil {
 		globalLogger.WithOptions(zap.AddCallerSkip(1)).Fatal(err)