Handle agent lock file on abrupt failures (#19)

cmd/viam-agent/main.go

@@ -104,12 +104,18 @@ func main() {
 	exitIfError(agent.InitPaths())
 
 	// use a lockfile to prevent running two agents on the same machine
-	pidFile, err := lockfile.New(filepath.Join(agent.ViamDirs["tmp"], "viam-agent.pid"))
+	pidFile, err := lockfile.New(filepath.Join(agent.ViamDirs["run"], "viam-agent.pid"))
 	exitIfError(errors.Wrap(err, "cannot init lock file"))
 	if err = pidFile.TryLock(); err != nil {
 		globalLogger.Error(errors.Wrapf(err, "cannot lock %s", pidFile))
 		if errors.Is(err, lockfile.ErrBusy) {
-			globalLogger.Fatal("Please terminate any other copies of viam-agent and try again.")
+			globalLogger.Debug("Retrying to lock file")
+
+			time.Sleep(2 * time.Second)
+
+			if err = pidFile.TryLock(); err != nil {
+				globalLogger.Fatal("Please terminate any other copies of viam-agent and try again.")
+			}
 		}
 	}
 	defer func() {
@@ -153,7 +159,7 @@ func main() {
 			if errors.Is(err, agent.ErrSubsystemDisabled) {
 				globalLogger.Warn("provisioning subsystem disabled, please manually update /etc/viam.json and connect to internet")
 			} else {
-				globalLogger.Error("could not start provisioning subsystem, please manually update /etc/viam.json and connect to internet")
+				globalLogger.Error(errors.Wrapf(err, "could not start provisioning subsystem, please manually update /etc/viam.json and connect to internet"))
 			}
 		}
 

subsystems/viamagent/viam-agent.service

@@ -9,6 +9,8 @@ Restart=always
 RestartSec=5
 User=root
 TimeoutSec=120
+RuntimeDirectory=viam
+RuntimeDirectoryMode=0755
 ExecStart=/opt/viam/bin/viam-agent --config /etc/viam.json
 FinalKillSignal=SIGQUIT
 

utils.go

@@ -29,6 +29,7 @@ func init() {
 	ViamDirs["cache"] = filepath.Join(ViamDirs["viam"], "cache")
 	ViamDirs["tmp"] = filepath.Join(ViamDirs["viam"], "tmp")
 	ViamDirs["etc"] = filepath.Join(ViamDirs["viam"], "etc")
+	ViamDirs["run"] = "/run/viam"
 }
 
 func InitPaths() error {