Merge pull request #102 from hmrc/BDOG-3097

BDOG-3097

README.md

@@ -28,24 +28,24 @@ If you'd prefer to carry out these steps manually, then follow these steps:
 
 **Linux Intel**
 ```shell
-curl -L -O https://github.com/hmrc/sm2/releases/download/v1.2.0/sm2-1.2.0-linux-intel.zip && unzip sm2-1.2.0-linux-intel.zip && rm sm2-1.2.0-linux-intel.zip && chmod +x sm2
+curl -L -O https://github.com/hmrc/sm2/releases/download/v2.0.0/sm2-2.0.0-linux-intel.zip && unzip sm2-2.0.0-linux-intel.zip && rm sm2-2.0.0-linux-intel.zip && chmod +x sm2
 ```
 
 **Linux Arm64**
 ```shell
-curl -L -O https://github.com/hmrc/sm2/releases/download/v1.2.0/sm2-1.2.0-linux-arm64.zip && unzip sm2-1.2.0-linux-arm64.zip && rm sm2-1.2.0-linux-arm64.zip && chmod +x sm2
+curl -L -O https://github.com/hmrc/sm2/releases/download/v2.0.0/sm2-2.0.0-linux-arm64.zip && unzip sm2-2.0.0-linux-arm64.zip && rm sm2-2.0.0-linux-arm64.zip && chmod +x sm2
 ```
 
 **OSX/Apple (latest M1/M2 cpus)**
 
 ```shell
-curl -L -O https://github.com/hmrc/sm2/releases/download/v1.2.0/sm2-1.2.0-apple-arm64.zip && unzip sm2-1.2.0-apple-arm64.zip && rm sm2-1.2.0-apple-arm64.zip && chmod +x sm2
+curl -L -O https://github.com/hmrc/sm2/releases/download/v2.0.0/sm2-2.0.0-apple-arm64.zip && unzip sm2-2.0.0-apple-arm64.zip && rm sm2-2.0.0-apple-arm64.zip && chmod +x sm2
 ```
 
 **OSX/Apple (older Intel cpus)**
 
 ```shell
-curl -L -O https://github.com/hmrc/sm2/releases/download/v1.2.0/sm2-1.2.0-apple-intel.zip && unzip sm2-1.2.0-apple-intel.zip && rm sm2-1.2.0-apple-intel.zip && chmod +x sm2
+curl -L -O https://github.com/hmrc/sm2/releases/download/v2.0.0/sm2-2.0.0-apple-intel.zip && unzip sm2-2.0.0-apple-intel.zip && rm sm2-2.0.0-apple-intel.zip && chmod +x sm2
 ```
 
 If everything has worked you should have an executable called `sm2`.
@@ -130,14 +130,6 @@ Alternatively you can start more than one service at once by typing multiple ser
 $ sm2 -start SERVICE_ONE SERVICE_TWO
 ```
 
-#### Starting a large group of services
-Starting a large group of services can overload the cpu of a machine and lead to services failing to start.
-If this happens use the following command to start the services at a slower pace.
-```shell
-$ sm2 --start LARGE_PROFILE_NAME --workers 1 --delay-seconds 5
-```
-The workers argument starts one service at a time and the DelaySeconds argument adds a 5 second delay inbetween services.
-
 ### Starting specific versions
 If you need to run a specific version of a service you can do so by adding a colon followed by the version number to the service name, e.g.
 ```shell
@@ -160,7 +152,7 @@ When starting more than one service, the `-r` flag only applies to the first ser
 | -noprogress    | Disabled the progress bars. Useful for scripting and automation.
 | -src           | Runs the service(s) from source instead of downloading the binary artifacts. Service manager will attempt to clone the repository and start the service using sbt start. Assumes the system has git configured and a working sbt installation.
 | -update-config | Updates workspace copy of service-manager from git. Will fail if there are uncommitted changes or if the config repo is not on the main branch.
-| -wait 120      | Waits a given number of seconds for the service to start before exiting.
+| -wait 120      | Waits a given number of seconds (default 30) for the service to respond to a healthcheck after startup.
 | -workers 4     | Sets the number of concurrent downloads (default 2). Can also be set via SM_WORKERS environment variable.
 
 ### Stopping services (-stop)
@@ -238,7 +230,7 @@ Sometimes a service will fail to start up. To help determine why, service manage
 Before doing anything else, it’s worth running service-manager’s self-checks to ensure it is installed correctly.
 ```shell
 $ sm2 -diagnostic
-version: 1.2.0
+version: 2.0.0
   build: ef49b60
 OS:		 OK (linux, amd64)
 JAVA:		 OK (11.0.17)

USERGUIDE.md

@@ -42,7 +42,7 @@ sm2 --start SERVICE_ONE SERVICE_TWO SERVICE_THREE
 | `--noprogress`    | Supresses the progress bars when downloading the service. Suitable for scripts etc.                                  |
 | `--offline`       | Starts services that are already without attempting to download the latest version                                   |
 | `--clean`         | Removes existing install, forcing a re-download                                                                      |
-| `--wait 20`       | Waits a specified number of seconds for all services to reach a healthy state                                        |
+| `--wait 20`       | Waits a specified number of seconds (default 30) for each service to reach a healthy state                                        |
 | `--appendArgs`    | A json map of extra args for services being started: `{"SERVICE_NAME":["-DFoo=Bar","SOMETHING"]}`                    |
 | `--workers 4`     | The number of services to download/start at the same time (default 2)                                                |
 | `--reverse-proxy` | Starts a reverse proxy                                                                                               |

cli/cli.go

@@ -50,9 +50,8 @@ type UserOption struct {
 	Verbose              bool                // shows extra logging
 	Version              bool                // prints sm2 version number
 	Verify               bool                // checks if a given service or profile is running
-	Wait                 int                 // waits given number of secs after starting services for then to respond to pings
+	Wait                 int                 // waits a given number of secs (default 30) after starting a service for it to respond to healthcheck
 	Workers              int                 // sets the number of concurrent downloads/service starts
-	DelaySeconds         int                 // sets the pause in seconds between starting services
 }
 
 func Parse(args []string) (*UserOption, error) {
@@ -197,9 +196,8 @@ func BuildFlagSet(opts *UserOption) *flag.FlagSet {
 	flagset.BoolVar(&opts.Verbose, "v", false, "enable verbose output")
 	flagset.BoolVar(&opts.Version, "version", false, "show the version of service-manager")
 	flagset.BoolVar(&opts.Verify, "verify", false, "for scripts, checks if a service/profile is running")
-	flagset.IntVar(&opts.Wait, "wait", 0, "used with --start, waits a specified number of seconds for the services to become available before exiting (use with --start)")
+	flagset.IntVar(&opts.Wait, "wait", 30, "used with --start, waits a specified number of seconds for each service to respond to a healthcheck, defaults to 30")
 	flagset.IntVar(&opts.Workers, "workers", defaultWorkers(), "how many services should be downloaded at the same time (use with --start)")
-	flagset.IntVar(&opts.DelaySeconds, "delay-seconds", 0, "how long to pause, in seconds, after starting a service before starting another")
 
 	return flagset
 }

install.sh

@@ -4,7 +4,7 @@ IFS=$'\n\t'
 
 TARGET_OS="linux"
 TARGET_ARCH="intel"
-TARGET_VERSION="1.2.0"
+TARGET_VERSION="2.0.0"
 
 if [[ "$(uname)" == "Darwin" ]]; then
     TARGET_OS="apple"

makefile

@@ -8,7 +8,7 @@
 ROOT_DIR:=$(shell dirname $(realpath $(lastword $(MAKEFILE_LIST))))
 
 BINARY := sm2
-VERSION := 1.2.0
+VERSION := 2.0.0
 BUILD := `git rev-parse HEAD`
 
 # Setup linker flags option for build that interoperate with variable names in src code

servicemanager/autocomplete.go

@@ -77,8 +77,7 @@ func dontComplete(previousWord string) bool {
 		"-ports",
 		"-search",
 		"-wait",
-		"-workers",
-		"-delay-seconds":
+		"-workers":
 		return true
 	}
 	return false

servicemanager/startservice.go

@@ -92,18 +92,24 @@ func (sm *ServiceManager) StartService(serviceAndVersion ServiceAndVersion) erro
 	state.HealthcheckUrl = healthcheckUrl
 	// and finally, we record out success
 	err = sm.Ledger.SaveStateFile(installDir, state)
-	sm.pauseTillHealthy(healthcheckUrl)
+	if err != nil {
+		sm.progress.update(serviceAndVersion.service, 0, "Failed")
+		return err
+	}
+	err = sm.pauseTillHealthy(healthcheckUrl)
 	return err
 }
 
-func (sm *ServiceManager) pauseTillHealthy(healthcheckUrl string) {
-	if sm.Commands.DelaySeconds > 0 {
-		count := 0
-		for count < sm.Commands.DelaySeconds*2 && !sm.CheckHealth(healthcheckUrl) {
-			count++
-			time.Sleep(500 * time.Millisecond)
+func (sm *ServiceManager) pauseTillHealthy(healthcheckUrl string) error {
+	count := 0
+	for count < sm.Commands.Wait*2 {
+		if sm.CheckHealth(healthcheckUrl) {
+			return nil
 		}
+		count++
+		time.Sleep(500 * time.Millisecond)
 	}
+	return fmt.Errorf("health check unsuccessful after %d seconds", sm.Commands.Wait)
 }
 
 func (sm *ServiceManager) installService(installDir string, serviceId string, group string, artifact string, version string) (ledger.InstallFile, error) {
@@ -347,17 +353,13 @@ func (sm *ServiceManager) asyncStart(services []ServiceAndVersion) {
 		fmt.Println("Starting 1 service")
 	} else {
 		workerPlural := "workers"
-		delay := ""
 
 		if sm.Commands.Workers == 1 {
 			workerPlural = "worker"
 		}
 
-		if sm.Commands.DelaySeconds != 0 {
-			delay = fmt.Sprintf(", with a delay of %d seconds", sm.Commands.DelaySeconds)
-		}
 
-		fmt.Printf("Starting %d services on %d %s%s\n", len(services), sm.Commands.Workers, workerPlural, delay)
+		fmt.Printf("Starting %d services on %d %s\n", len(services), sm.Commands.Workers, workerPlural)
 	}
 
 	// start up a number of workers (controlled by --workers param)
@@ -376,13 +378,8 @@ func (sm *ServiceManager) asyncStart(services []ServiceAndVersion) {
 	// this could be way better, wait groups, or force a final paint or something??
 	time.Sleep(time.Millisecond)
 
-	if sm.Commands.Wait > 0 {
-		fmt.Printf("Waiting %d secs for all services to start.", sm.Commands.Wait)
-		sm.Await(services, sm.Commands.Wait)
-	}
-
 	// if anything has failed to start, report why
-	if len(sm.progress.errors) > 0 && !sm.progress.noProgress {
+	if len(sm.progress.errors) > 0 {
 		fmt.Println("\n\033[1;31mSome services failed to start:\033[0m")
 		for k, v := range sm.progress.errors {
 			fmt.Printf("  %s: %s\n", k, v.Error())

servicemanager/status.go

@@ -49,7 +49,7 @@ func (sm *ServiceManager) PrintStatus() {
 
 		longestServiceName := getLongestServiceName(append(statuses, unmanaged...))
 		printTable(statuses, termWidth, longestServiceName, os.Stdout)
-		printHelpIfRequired(statuses, sm.Commands.DelaySeconds)
+		printHelpIfRequired(statuses)
 
 		if len(unmanaged) > 0 {
 			fmt.Print("\n\033[34mAlso, the following processes are running which occupy ports of services\n")
@@ -332,7 +332,7 @@ func printUnmanagedTable(statuses []serviceStatus, maxWidth int, longestServiceN
 	fmt.Fprint(out, border)
 }
 
-func printHelpIfRequired(statuses []serviceStatus, delay int) {
+func printHelpIfRequired(statuses []serviceStatus) {
 	for _, status := range statuses {
 		if status.health == FAIL && status.service != "MONGO" {
 			fmt.Print("\n\033[1;31mOne or more services have failed to start.\033[0m\n")
@@ -342,14 +342,6 @@ func printHelpIfRequired(statuses []serviceStatus, delay int) {
 			fmt.Print("Alternatively, you can remove them from this list by using:\n")
 			fmt.Print("  sm2 -prune\n\n")
 
-			if delay == 0 && len(statuses) >= 10 { // not already using --delay-seconds
-				fmt.Println("Note: If you're starting a profile that contains a lot of services,")
-				fmt.Println("try using `--delay-seconds 5` to add a 5 second delay after starting")
-				fmt.Println("each service. This will help prevent your CPU getting overloaded,")
-				fmt.Println("which can cause the services to take too long to respond to the healthcheck.")
-				fmt.Println("See `sm2 --help` for more information.")
-			}
-
 			return
 		}
 	}