arthur and config testing and fixes

merlin/README.md

@@ -20,6 +20,7 @@ This folder contains the code for merlin (which does the actual syncing) and art
 - [ ] wget
 
 ### TODO
+- [ ] ensure that the proper permissions (file mode, group, user) are used
 - [ ] detect if an rsync process is stuck (\*\*)
 - [ ] place each rsync process in a separate cgroup (\*\*\*)
 

merlin/arthur/arthur.go

@@ -1,13 +1,13 @@
 package arthur
 
 import (
-	"bytes"
 	"errors"
 	"fmt"
-	"io"
+	"io/ioutil"
 	"net"
 	"os"
 	"path/filepath"
+	"sort"
 	"strings"
 	"text/tabwriter"
 	"time"
@@ -17,68 +17,83 @@ import (
 	"git.csclub.uwaterloo.ca/public/merlin/sync"
 )
 
-// get repo by name function in config
+// Reads and parses the message sent over the accepted connection
+func GetCommand(conn net.Conn) (command, repoName string) {
+	command = ""
+	repoName = ""
 
-func GetAndRunCommand(conn net.Conn) (newSync bool) {
-	newSync = false
-	defer conn.Close()
-
-	var buf bytes.Buffer
-	_, err := io.Copy(&buf, conn)
+	buf, err := ioutil.ReadAll(conn)
 	if err != nil {
 		logger.ErrLog(err.Error())
 		return
 	}
 
-	command := buf.String()
-	args := strings.Split(command, ":")
-	respondAndLogErr := func(msg string) {
-		logger.OutLog(msg)
-		conn.Write([]byte(msg))
+	args := strings.Split(string(buf), ":")
+	if len(args) >= 1 {
+		command = args[0]
 	}
-
-	if args[0] == "status" {
-		status := tabwriter.NewWriter(conn, 5, 5, 5, ' ', 0)
-		fmt.Fprintf(status, "Repository\tLast Synced\tNext Expected Sync\n")
-
-		// for time formating see https://pkg.go.dev/time#pkg-constants
-		for name, repo := range config.RepoMap {
-			fmt.Fprintf(status, "%s\t%s\t%s\n",
-				name,
-				time.Unix(repo.State.LastAttemptStartTime, 0).Format(time.RFC1123),
-				time.Unix(repo.State.LastAttemptRunTime+int64(repo.Frequency), 0).Format(time.RFC1123),
-			)
-		}
-
-		status.Flush()
-	} else if args[0] == "sync" {
-		if len(args) != 2 {
-			respondAndLogErr("Could not parse sync command, forced sync fails.")
-			return
-		}
-
-		if repo, inMap := config.RepoMap[args[1]]; inMap {
-			logger.OutLog("Attempting to force sync of " + repo.Name)
-			if sync.SyncIfPossible(repo) {
-				conn.Write([]byte("Forced sync for " + repo.Name))
-				newSync = true
-			} else {
-				respondAndLogErr("Cannot force sync: " + repo.Name + ", already syncing.")
-			}
-		} else {
-			respondAndLogErr(args[1] + " is not tracked so cannot sync")
-		}
-	} else {
-		respondAndLogErr("Received unrecognized command: " + command)
+	if len(args) >= 2 {
+		repoName = args[1]
 	}
 	return
 }
 
-func UnixSockListener(connChan chan net.Conn, stopLisChan chan struct{}) {
+func SendAndLog(conn net.Conn, msg string) {
+	logger.OutLog(msg)
+	conn.Write([]byte(msg))
+}
+
+func SendStatus(conn net.Conn) {
+	status := tabwriter.NewWriter(conn, 5, 5, 5, ' ', 0)
+	fmt.Fprintf(status, "Repository\tLast Synced\tNext Expected Sync\tRunning\n")
+
+	keys := make([]string, 0)
+	for name, _ := range config.RepoMap {
+		keys = append(keys, name)
+	}
+	sort.Strings(keys)
+
+	// for other ways to format the time see: https://pkg.go.dev/time#pkg-constants
+	for _, name := range keys {
+		repo := config.RepoMap[name]
+		lastSync := repo.State.LastAttemptStartTime
+		nextSync := lastSync + int64(repo.Frequency)
+
+		fmt.Fprintf(status, "%s\t%s\t%s\t%t\n",
+			name,
+			time.Unix(lastSync, 0).Format(time.RFC1123),
+			time.Unix(nextSync, 0).Format(time.RFC1123),
+			repo.State.IsRunning,
+		)
+	}
+
+	status.Flush()
+}
+
+// Attempt to force the sync of the repo
+func ForceSync(conn net.Conn, repoName string) (newSync bool) {
+	newSync = false
+
+	// TODO: send repoName and every key in RepoMap to lowercase
+	if repo, isInMap := config.RepoMap[repoName]; isInMap {
+		logger.OutLog("Attempting to force sync of " + repoName)
+		if sync.SyncIfPossible(repo) {
+			conn.Write([]byte("Forced sync for " + repoName))
+			newSync = true
+		} else {
+			SendAndLog(conn, "Could not force sync: "+repoName+" is already syncing.")
+		}
+	} else {
+		SendAndLog(conn, repoName+" is not tracked so cannot sync")
+	}
+	return
+}
+
+func StartListener(connChan chan net.Conn, stopLisChan chan struct{}) {
 	sockpath := config.Conf.SockPath
-	// must remove cfg.SockPath otherwise get "bind: address already in use"
+	// must remove old cfg.SockPath otherwise get "bind: address already in use"
 	if filepath.Ext(sockpath) != ".sock" {
-		panic(fmt.Errorf("Socket file must end with .sock"))
+		panic(fmt.Errorf("socket file must end with .sock"))
 	} else if _, err := os.Stat(sockpath); err == nil {
 		if err := os.Remove(sockpath); err != nil {
 			panic(err)
@@ -95,10 +110,10 @@ func UnixSockListener(connChan chan net.Conn, stopLisChan chan struct{}) {
 
 	go func() {
 		for {
-			// will exit when ear is closed
+			// Attempting to accept on a closed net.Listener will return a non-temporary error
 			conn, err := ear.Accept()
 			if err != nil {
-				if ne, ok := err.(net.Error); ok && ne.Temporary() {
+				if netErr, isTemp := err.(net.Error); isTemp && netErr.Temporary() {
 					logger.ErrLog("Accepted socket error: " + err.Error())
 					continue
 				}
@@ -109,6 +124,7 @@ func UnixSockListener(connChan chan net.Conn, stopLisChan chan struct{}) {
 		}
 	}()
 
+	// TODO: check handling of multiple SIGHUP
 	<-stopLisChan
 	ear.Close()
 }

merlin/arthur/arthur_test.go

@@ -1 +1,283 @@
 package arthur
+
+import (
+	"io/ioutil"
+	"net"
+	"os"
+	"testing"
+	"time"
+
+	"git.csclub.uwaterloo.ca/public/merlin/config"
+	"git.csclub.uwaterloo.ca/public/merlin/logger"
+)
+
+func TestStatusCommand(t *testing.T) {
+	r, w := net.Pipe()
+
+	go func() {
+		// will only finish write when EOF is sent
+		// only way to send EOF is to close
+		w.Write([]byte("status"))
+		w.Close()
+	}()
+	command, repoName := GetCommand(r)
+	if command != "status" {
+		t.Errorf("command for status should be \"status\", got " + command)
+	} else if repoName != "" {
+		t.Errorf("status should return an empty string for the repoName, got " + repoName)
+	}
+}
+
+func TestSyncCommand(t *testing.T) {
+	r, w := net.Pipe()
+
+	go func() {
+		w.Write([]byte("sync:ubuntu"))
+		w.Close()
+	}()
+	command, repoName := GetCommand(r)
+	r.Close()
+	if command != "sync" {
+		t.Errorf("command for sync:ubuntu should be \"sync\", got " + command)
+	} else if repoName != "ubuntu" {
+		t.Errorf("name of repo for sync:ubuntu should be \"ubuntu\", got " + repoName)
+	}
+}
+
+func TestSendStatus(t *testing.T) {
+	saveRepoMap := config.RepoMap
+	defer func() {
+		config.RepoMap = saveRepoMap
+	}()
+
+	repoMap := make(map[string]*config.Repo)
+	repoMap["eeeee"] = &config.Repo{
+		Frequency: 30 * 86400,
+		State: config.RepoState{
+			IsRunning:            true,
+			LastAttemptStartTime: 1600000000,
+		},
+	}
+	repoMap["alinux"] = &config.Repo{
+		Frequency: 7*86400 + 3,
+		State: config.RepoState{
+			IsRunning:            true,
+			LastAttemptStartTime: 1620000000,
+		},
+	}
+	repoMap["lnux"] = &config.Repo{
+		Frequency: 86400,
+		State: config.RepoState{
+			IsRunning:            false,
+			LastAttemptStartTime: 1640000000,
+		},
+	}
+	config.RepoMap = repoMap
+
+	r, w := net.Pipe()
+	go func() {
+		SendStatus(w)
+		w.Close()
+	}()
+	msg, err := ioutil.ReadAll(r)
+	r.Close()
+	if err != nil {
+		t.Errorf(err.Error())
+	}
+	expected := `Repository     Last Synced                       Next Expected Sync                Running
+alinux         Sun, 02 May 2021 20:00:00 EDT     Sun, 09 May 2021 20:00:03 EDT     true
+eeeee          Sun, 13 Sep 2020 08:26:40 EDT     Tue, 13 Oct 2020 08:26:40 EDT     true
+lnux           Mon, 20 Dec 2021 06:33:20 EST     Tue, 21 Dec 2021 06:33:20 EST     false
+`
+	if expected != string(msg) {
+		t.Errorf("Expected:\n" + expected + "\nGot:\n" + string(msg))
+	}
+}
+
+func TestForceSync(t *testing.T) {
+	saveRepos := config.Repos
+	saveRepoMap := config.RepoMap
+	doneChan := make(chan config.SyncResult)
+	defer func() {
+		config.Repos = saveRepos
+		config.RepoMap = saveRepoMap
+		close(doneChan)
+	}()
+
+	// Part 1: run a dummy sync
+	repo := config.Repo{
+		Name:      "nux",
+		SyncType:  "csc-sync-dummy",
+		Frequency: 7 * 86400,
+		MaxTime:   30,
+		Logger:    logger.NewLogger("nux", "/tmp/merlin_force_sync_test_logs"),
+		StateFile: "/tmp/merlin_force_sync_test_state",
+		DoneChan:  doneChan,
+		State: config.RepoState{
+			IsRunning:            false,
+			LastAttemptStartTime: 0,
+			LastAttemptRunTime:   0,
+			LastAttemptExit:      config.NOT_RUN_YET,
+		},
+	}
+	config.Repos = nil
+	config.Repos = append(config.Repos, &repo)
+	config.RepoMap = make(map[string]*config.Repo)
+	config.RepoMap["nux"] = &repo
+
+	r, w := net.Pipe()
+	go func() {
+		if !ForceSync(w, "nux") {
+			t.Errorf("Sync for nux did not start")
+		}
+		w.Close()
+	}()
+	msg, err := ioutil.ReadAll(r)
+	r.Close()
+	if err != nil {
+		t.Errorf(err.Error())
+	}
+	expected := "Forced sync for nux"
+	if expected != string(msg) {
+		t.Errorf("Expected:\n" + expected + "\nGot:\n" + string(msg))
+	}
+
+	select {
+	case result := <-doneChan:
+		if result.Exit != config.SUCCESS {
+			t.Errorf("Sync should exit with SUCCESS, got %d", result.Exit)
+		}
+	case <-time.After(3 * time.Second):
+		t.Errorf("Dummy sync should be done in 1 second, waited 3 seconds")
+	}
+
+	// Part 2: attempt the same thing but with repo.State.IsRunning = true
+	r, w = net.Pipe()
+	go func() {
+		if ForceSync(w, "nux") {
+			t.Errorf("Sync for nux should not have started")
+		}
+		w.Close()
+	}()
+	msg, err = ioutil.ReadAll(r)
+	r.Close()
+	if err != nil {
+		t.Errorf(err.Error())
+	}
+	expected = "Could not force sync: nux is already syncing."
+	if expected != string(msg) {
+		t.Errorf("Expected:\n" + expected + "\nGot:\n" + string(msg))
+	}
+
+	select {
+	case <-doneChan:
+		t.Errorf("Sync for nux should not have been started")
+	case <-time.After(2 * time.Second):
+	}
+
+	// Part 3: attempt a force sync with a repo that does not exist
+	r, w = net.Pipe()
+	go func() {
+		if ForceSync(w, "nixx") {
+			t.Errorf("Sync for nixx should not have started")
+		}
+		w.Close()
+	}()
+	msg, err = ioutil.ReadAll(r)
+	r.Close()
+	if err != nil {
+		t.Errorf(err.Error())
+	}
+	expected = "nixx is not tracked so cannot sync"
+	if expected != string(msg) {
+		t.Errorf("Expected:\n" + expected + "\nGot:\n" + string(msg))
+	}
+}
+
+func TestStartListener(t *testing.T) {
+	saveConf := config.Conf
+	connChan := make(chan net.Conn)
+	stopLisChan := make(chan struct{})
+	wait := make(chan struct{})
+	defer func() {
+		config.Conf = saveConf
+		close(connChan)
+		close(stopLisChan)
+	}()
+	config.Conf = config.Config{
+		SockPath: "/tmp/merlin_listener_test.sock",
+	}
+
+	// Test 1: check that closing/sending something to stopLisChan will stop the listener
+	// 		   and that a new listener can be created after stopping the old one
+	go func() {
+		StartListener(connChan, stopLisChan)
+		wait <- struct{}{}
+	}()
+	stopLisChan <- struct{}{}
+	select {
+	case <-wait:
+	case <-time.After(3 * time.Second):
+		t.Errorf("StartListener should stop when struct{}{} is sent to stopLisChan")
+	}
+
+	go func() {
+		StartListener(connChan, stopLisChan)
+		wait <- struct{}{}
+	}()
+	close(stopLisChan)
+	select {
+	case <-wait:
+	case <-time.After(3 * time.Second):
+		t.Errorf("StartListener should stop when stopLisChan is closed")
+	}
+	close(wait)
+
+	// Test 2: check that connections can be made to the unix socket
+	// 		   this test does not appear to be very stable (I think there is a race condition somewhere)
+	stopLisChan = make(chan struct{})
+	go StartListener(connChan, stopLisChan)
+	waitForMsg := func(expected string) {
+		select {
+		case conn := <-connChan:
+			msg, err := ioutil.ReadAll(conn)
+			if err != nil {
+				t.Errorf(err.Error())
+			} else if expected != string(msg) {
+				t.Errorf("Message expected was " + expected + " got " + string(msg))
+			}
+			conn.Close()
+		case <-time.After(3 * time.Second):
+			t.Errorf("StartListener should stop when struct{}{} is sent to stopLisChan")
+		}
+	}
+	sendMsg := func(msg string) {
+		<-time.After(500 * time.Millisecond)
+		send, err := net.Dial("unix", "/tmp/merlin_listener_test.sock")
+		if err != nil {
+			panic(err)
+		}
+		_, err = send.Write([]byte(msg))
+		if err != nil {
+			t.Errorf(err.Error())
+		}
+		send.Close()
+	}
+	go func() {
+		waitForMsg("status")
+	}()
+	sendMsg("status")
+
+	go func() {
+		waitForMsg("sync:uuunix")
+	}()
+	sendMsg("sync:uuunix")
+
+	go func() {
+		waitForMsg("$UPz2L2yWsE^UY8iG9JX@^dBb@5yb*")
+	}()
+	sendMsg("$UPz2L2yWsE^UY8iG9JX@^dBb@5yb*")
+
+	// unsure why I can't put this in the defer
+	os.Remove("/tmp/merlin_listener_test.sock")
+}

merlin/config/config.go

@@ -2,6 +2,7 @@ package config
 
 import (
 	"os"
+	"path/filepath"
 
 	"gopkg.in/ini.v1"
 
@@ -18,17 +19,18 @@ const (
 	TEN_MINUTELY  = 600
 	FIVE_MINUTELY = 300
 
-	CONFIG_PATH = "merlin-config.ini"
-
-	DEFAULT_MAX_JOBS        = 6
-	DEFAULT_MAX_TIME        = DAILY / 4
-	DEFAULT_DOWNLOAD_DIR    = "/mirror/root"
-	DEFAULT_PASSWORD_DIR    = "/home/mirror/passwords"
-	DEFAULT_STATE_DIR       = "/home/mirror/merlin/states"
-	DEFAULT_LOG_DIR         = "/home/mirror/merlin/logs"
-	DEFAULT_RSYNC_LOG_DIR   = "/home/mirror/merlin/logs-rsync"
-	DEFAULT_ZFSSYNC_LOG_DIR = "/home/mirror/merlin/logs-zfssync"
-	DEFAULT_SOCK_PATH       = "/run/merlin.sock"
+	// could change this into a default_config
+	DEFAULT_MAX_JOBS         = 6
+	DEFAULT_MAX_TIME         = DAILY / 4
+	DEFAULT_SYNC_TYPE        = "csc-sync-standard"
+	DEFAULT_FREQUENCY_STRING = "by-hourly"
+	DEFAULT_DOWNLOAD_DIR     = "/mirror/root"
+	DEFAULT_PASSWORD_DIR     = "/home/mirror/passwords"
+	DEFAULT_STATE_DIR        = "/home/mirror/merlin/states"
+	DEFAULT_LOG_DIR          = "/home/mirror/merlin/logs"
+	DEFAULT_RSYNC_LOG_DIR    = "/home/mirror/merlin/logs-rsync"
+	DEFAULT_ZFSSYNC_LOG_DIR  = "/home/mirror/merlin/logs-zfssync"
+	DEFAULT_SOCK_PATH        = "/run/merlin.sock"
 )
 
 var frequencies = map[string]int{
@@ -62,11 +64,11 @@ type Config struct {
 	IPv4Address string `ini:"ipv4_address"`
 	IPv6Address string `ini:"ipv6_address"`
 	// the default sync type
-	SyncType string `ini:"default_sync_type"`
-	// the default frequency string for the repos
-	FrequencyStr string `ini:"default_frequency"`
-	// the default MaxTime for each repo
-	MaxTime int `ini:"default_max_time"`
+	DefaultSyncType string `ini:"default_sync_type"`
+	// the default frequency string
+	DefaultFrequencyStr string `ini:"default_frequency"`
+	// the default MaxTime
+	DefaultMaxTime int `ini:"default_max_time"`
 	// the directory where rsync should download files
 	DownloadDir string `ini:"download_dir"`
 	// the directory where rsync passwords are stored
@@ -74,15 +76,16 @@ type Config struct {
 	// the directory where the state of each repo sync is saved
 	StateDir string `ini:"states_dir"`
 	// the directory where merlin will store the general logs for each repo
-	LoggerDir string `ini:"log_dir"`
+	RepoLogDir string `ini:"repo_logs_dir"`
 	// the directory to store the rsync logs for each repo
-	RsyncLogDir string `ini:"rsync_log_dir"`
+	RsyncLogDir string `ini:"rsync_logs_dir"`
 	// the directory to store the zfssync logs for each repo
-	ZfssyncLogDir string `ini:"zfssync_log_dir"`
+	ZfssyncLogDir string `ini:"zfssync_logs_dir"`
 	// the Unix socket path which arthur will use to communicate with us
 	SockPath string `ini:"sock_path"`
 }
 
+// make it more clear when full path should be used vs when just the file name is needed
 type Repo struct {
 	// the name of this repo
 	Name string `ini:"-"`
@@ -106,18 +109,20 @@ type Repo struct {
 	RsyncUser string `ini:"rsync_user"`
 	// the file storing the password for rsync (optional)
 	PasswordFile string `ini:"password_file"`
-	// the file for general logging of this repo
-	LoggerFile string `ini:"log_file"`
+	// the file storing the repo sync state (used to override default)
+	StateFile string `ini:"state_file"`
+	// the full file path for general logging of this repo (used to override default)
+	RepoLogFile string `ini:"repo_log_file"`
 	// a reference to the general logger
 	Logger *logger.Logger `ini:"-"`
-	// the file for logging this repo's rsync
+	// the full file path for logging this repo's rsync (used to override default)
 	RsyncLogFile string `ini:"rsync_log_file"`
-	// the file for logging this repo's zfssync
+	// the full file path for logging this repo's zfssync (used to override default)
 	ZfssyncLogFile string `ini:"zfssync_log_file"`
 	// the repo will write its name and status in a Result struct to DoneChan
 	// when it has finished a job (shared by all repos)
 	DoneChan chan<- SyncResult `ini:"-"`
-	// the repo should stop syncing if StopChan is closed (shared by all repos)
+	// repos should stop syncing if StopChan is closed (shared by all repos)
 	StopChan chan struct{} `ini:"-"`
 	// a struct that stores the repo's status
 	State RepoState `ini:"-"`
@@ -145,32 +150,34 @@ var (
 // GetConfig reads the config from a JSON file, initializes default values,
 // and initializes the non-configurable fields of each repo.
 // It returns a Config.
-func LoadConfig(doneChan chan SyncResult, stopChan chan struct{}) {
+func LoadConfig(configPath string, doneChan chan SyncResult, stopChan chan struct{}) {
 	// set default values then load config from file
 	newConf := Config{
-		MaxJobs:       DEFAULT_MAX_JOBS,
-		MaxTime:       DEFAULT_MAX_TIME,
-		PasswordDir:   DEFAULT_PASSWORD_DIR,
-		DownloadDir:   DEFAULT_DOWNLOAD_DIR,
-		StateDir:      DEFAULT_STATE_DIR,
-		LoggerDir:     DEFAULT_LOG_DIR,
-		RsyncLogDir:   DEFAULT_RSYNC_LOG_DIR,
-		ZfssyncLogDir: DEFAULT_ZFSSYNC_LOG_DIR,
-		SockPath:      DEFAULT_SOCK_PATH,
+		MaxJobs:             DEFAULT_MAX_JOBS,
+		DefaultSyncType:     DEFAULT_SYNC_TYPE,
+		DefaultFrequencyStr: DEFAULT_FREQUENCY_STRING,
+		DefaultMaxTime:      DEFAULT_MAX_TIME,
+		PasswordDir:         DEFAULT_PASSWORD_DIR,
+		DownloadDir:         DEFAULT_DOWNLOAD_DIR,
+		StateDir:            DEFAULT_STATE_DIR,
+		RepoLogDir:          DEFAULT_LOG_DIR,
+		RsyncLogDir:         DEFAULT_RSYNC_LOG_DIR,
+		ZfssyncLogDir:       DEFAULT_ZFSSYNC_LOG_DIR,
+		SockPath:            DEFAULT_SOCK_PATH,
 	}
-	iniInfo, err := ini.Load(CONFIG_PATH)
+	iniInfo, err := ini.Load(configPath)
 	panicIfErr(err)
 	err = iniInfo.MapTo(&newConf)
 	panicIfErr(err)
 
 	// check config for major errors
-	for _, dir := range []string{Conf.StateDir, Conf.LoggerDir, Conf.RsyncLogDir, Conf.ZfssyncLogDir} {
+	for _, dir := range []string{newConf.StateDir, newConf.RepoLogDir, newConf.RsyncLogDir, newConf.ZfssyncLogDir} {
 		err := os.MkdirAll(dir, 0755)
 		panicIfErr(err)
 	}
-	if Conf.IPv4Address == "" {
+	if newConf.IPv4Address == "" {
 		panic("Missing IPv4 address from config")
-	} else if Conf.IPv6Address == "" {
+	} else if newConf.IPv6Address == "" {
 		panic("Missing IPv6 address from config")
 	}
 
@@ -182,26 +189,32 @@ func LoadConfig(doneChan chan SyncResult, stopChan chan struct{}) {
 		}
 
 		// set the default values for the repo then load from file
+		// TODO: check if local_dir and repoName are always the same value
+		// TODO: check to ensure that every Repo.Name is unique (may already be done by ini)
 		repo := Repo{
 			Name:           repoName,
-			SyncType:       Conf.SyncType,
-			FrequencyStr:   Conf.FrequencyStr,
-			MaxTime:        Conf.MaxTime,
-			LoggerFile:     Conf.LoggerDir + "/" + repoName + ".log",
-			RsyncLogFile:   Conf.RsyncLogDir + "/" + repoName + "-rsync.log",
-			ZfssyncLogFile: Conf.ZfssyncLogDir + "/" + repoName + "-zfssync.log",
+			SyncType:       newConf.DefaultSyncType,
+			FrequencyStr:   newConf.DefaultFrequencyStr,
+			MaxTime:        newConf.DefaultMaxTime,
+			StateFile:      filepath.Join(newConf.StateDir, repoName),
+			RepoLogFile:    filepath.Join(newConf.RepoLogDir, repoName) + ".log",
+			RsyncLogFile:   filepath.Join(newConf.RsyncLogDir, repoName) + "-rsync.log",
+			ZfssyncLogFile: filepath.Join(newConf.ZfssyncLogDir, repoName) + "-zfssync.log",
 			DoneChan:       doneChan,
 			StopChan:       stopChan,
 		}
 		err := section.MapTo(&repo)
 		panicIfErr(err)
 
+		// TODO: ensure that the parent dirs to the file also exist when touching
+		// or just remove the ability to override
 		touchFiles(
-			repo.LoggerFile,
+			repo.StateFile,
+			repo.RepoLogFile,
 			repo.RsyncLogFile,
 			repo.ZfssyncLogFile,
 		)
-		repo.Logger = logger.NewLogger(repo.Name, repo.LoggerFile)
+		repo.Logger = logger.NewLogger(repo.Name, repo.RepoLogFile)
 		repo.Frequency = frequencies[repo.FrequencyStr]
 		if repo.SyncType == "" {
 			panic("Missing sync type from " + repo.Name)
@@ -215,16 +228,9 @@ func LoadConfig(doneChan chan SyncResult, stopChan chan struct{}) {
 			LastAttemptRunTime:   0,
 			LastAttemptExit:      NOT_RUN_YET,
 		}
-
-		// create the state file if it does not exist, otherwise load it from existing file
-		repoStateFile := Conf.StateDir + "/" + repo.Name
-		if _, err := os.Stat(repoStateFile); err != nil {
-			touchFile(repoStateFile)
-			repo.SaveState()
-		} else {
-			err := ini.MapTo(&repo.State, repoStateFile)
-			panicIfErr(err)
-		}
+		err = ini.MapTo(&repo.State, repo.StateFile)
+		panicIfErr(err)
+		repo.SaveState()
 
 		newRepos = append(newRepos, &repo)
 	}
@@ -235,6 +241,7 @@ func LoadConfig(doneChan chan SyncResult, stopChan chan struct{}) {
 
 	Conf = newConf
 	Repos = newRepos
+	RepoMap = make(map[string]*Repo)
 	for _, repo := range Repos {
 		RepoMap[repo.Name] = repo
 	}
@@ -242,17 +249,17 @@ func LoadConfig(doneChan chan SyncResult, stopChan chan struct{}) {
 
 // save the current state of the repo to a file
 func (repo *Repo) SaveState() {
-	repo.Logger.Debug("Saving state")
+	// repo.Logger.Debug("Saving state")
 	state_cfg := ini.Empty()
 	if err := ini.ReflectFrom(state_cfg, &repo.State); err != nil {
 		repo.Logger.Error(err.Error())
 	}
-	file, err := os.OpenFile(Conf.StateDir+"/"+repo.Name, os.O_RDWR|os.O_CREATE, 0644)
+	file, err := os.OpenFile(repo.StateFile, os.O_RDWR|os.O_CREATE, 0644)
 	if err != nil {
 		repo.Logger.Error(err.Error())
 	}
 	if _, err := state_cfg.WriteTo(file); err != nil {
 		repo.Logger.Error(err.Error())
 	}
-	repo.Logger.Debug("Saved state")
+	// repo.Logger.Debug("Saved state")
 }

merlin/config/config_test.go

@@ -1 +1,122 @@
 package config
+
+import (
+	"errors"
+	"os"
+	"reflect"
+	"testing"
+
+	"github.com/davecgh/go-spew/spew"
+)
+
+func TestTouchFiles(t *testing.T) {
+	files := []string{
+		"/tmp/merlin_touch_test_1",
+		"/tmp/merlin_touch_test_2",
+		"/tmp/merlin_touch_test_3",
+	}
+	touchFiles(files[0], files[1], files[2])
+	for _, file := range files {
+		if _, err := os.Stat(file); err != nil {
+			t.Errorf(err.Error())
+		} else if err := os.Remove(file); err != nil {
+			t.Errorf(err.Error())
+		}
+	}
+}
+
+func TestPanicIfErr(t *testing.T) {
+	panicIfErr(nil)
+	defer func() { recover() }()
+	panicIfErr(errors.New("AAAAAAAAAA"))
+	t.Errorf("panicIfErr should have panicked")
+}
+
+func TestLoadConfig(t *testing.T) {
+	doneChan := make(chan SyncResult)
+	stopChan := make(chan struct{})
+	LoadConfig("config_test.ini", doneChan, stopChan)
+	// TODO: Fill out parts not part of the ini or state file
+	expectedConfig := Config{
+		MaxJobs:             6,
+		IPv4Address:         "129.97.134.129",
+		IPv6Address:         "2620:101:f000:4901:c5c::129",
+		DefaultSyncType:     "csc-sync-standard",
+		DefaultFrequencyStr: "daily",
+		DefaultMaxTime:      1000,
+		DownloadDir:         "/tmp/test-mirror",
+		PasswordDir:         "/home/mirror/passwords",
+		StateDir:            "test_files",
+		RepoLogDir:          "test_files/logs",
+		RsyncLogDir:         "test_files/rsync",
+		ZfssyncLogDir:       "test_files/zfssync",
+		SockPath:            "test_files/test.sock",
+	}
+	expectedRepo1 := Repo{
+		Name:           "eelinux",
+		SyncType:       "csc-sync-nonstandard",
+		FrequencyStr:   "tri-hourly",
+		Frequency:      10800,
+		MaxTime:        2000,
+		LocalDir:       "eelinux",
+		RsyncHost:      "rsync.releases.eelinux.ca",
+		RsyncDir:       "releases",
+		StateFile:      "test_files/eeelinux",
+		RepoLogFile:    "test_files/logs/eelinux.log",
+		Logger:         Repos[0].Logger,
+		RsyncLogFile:   "test_files/rsync/eelinux.log",
+		ZfssyncLogFile: "test_files/zfssync/eelinux.log",
+		DoneChan:       doneChan,
+		StopChan:       stopChan,
+		State: RepoState{
+			IsRunning:            false,
+			LastAttemptStartTime: 1600000000,
+			LastAttemptRunTime:   100,
+			LastAttemptExit:      1,
+		},
+	}
+	expectedRepo2 := Repo{
+		Name:           "yoland",
+		SyncType:       "csc-sync-standard",
+		FrequencyStr:   "daily",
+		Frequency:      86400,
+		MaxTime:        1000,
+		LocalDir:       "yoland-releases",
+		RsyncHost:      "rsync.releases.yoland.io",
+		RsyncDir:       "releases",
+		StateFile:      "test_files/yoland",
+		RepoLogFile:    "test_files/logs/yoland.log",
+		Logger:         Repos[1].Logger,
+		RsyncLogFile:   "test_files/rsync/yoland-rsync.log",
+		ZfssyncLogFile: "test_files/zfssync/yoland-zfssync.log",
+		DoneChan:       doneChan,
+		StopChan:       stopChan,
+		State: RepoState{
+			IsRunning:            false,
+			LastAttemptStartTime: 0,
+			LastAttemptRunTime:   0,
+			LastAttemptExit:      0,
+		},
+	}
+
+	if !reflect.DeepEqual(expectedConfig, Conf) {
+		t.Errorf("Config loaded does not match expected config")
+		spew.Dump(expectedConfig)
+		spew.Dump(Conf)
+	}
+	if !reflect.DeepEqual(expectedRepo1, *Repos[0]) {
+		t.Errorf("The eelinux repo loaded does not match the exected repo config")
+		spew.Dump(expectedRepo1)
+		spew.Dump(*Repos[0])
+	}
+	if !reflect.DeepEqual(expectedRepo2, *Repos[1]) {
+		t.Errorf("The yoland repo loaded does not match the exected repo config")
+		spew.Dump(expectedRepo2)
+		spew.Dump(*Repos[1])
+	}
+
+	os.Remove("test_files/yoland")
+	os.RemoveAll("test_files/logs")
+	os.RemoveAll("test_files/rsync")
+	os.RemoveAll("test_files/zfssync")
+}

merlin/config/config_test.ini

@@ -0,0 +1,39 @@
+; default values are commented out
+
+; max_jobs = 6
+ipv4_address = 129.97.134.129
+ipv6_address = 2620:101:f000:4901:c5c::129
+; default_sync_type = csc-sync-standard
+default_frequency = daily
+default_max_time = 1000
+download_dir = /tmp/test-mirror
+; password_dir = /home/mirror/passwords
+states_dir = test_files
+repo_logs_dir = test_files/logs
+rsync_logs_dir = test_files/rsync
+zfssync_logs_dir = test_files/zfssync
+sock_path = test_files/test.sock
+
+[eelinux]
+sync_type = csc-sync-nonstandard
+frequency = tri-hourly
+max_time = 2000
+local_dir = eelinux
+rsync_host = rsync.releases.eelinux.ca
+rsync_dir = releases
+state_file = test_files/eeelinux
+repo_log_file = test_files/logs/eelinux.log
+rsync_log_file = test_files/rsync/eelinux.log
+zfssync_log_file = test_files/zfssync/eelinux.log
+
+[yoland]
+; sync_type = csc-sync-standard
+; frequency = daily
+; max_time = 1000
+local_dir = yoland-releases
+rsync_host = rsync.releases.yoland.io
+rsync_dir = releases
+; state_file = test_files/yoland
+; repo_log_file = test_files/logs/yoland.log
+; rsync_log_file = test_files/rsync/yoland-rsync.log
+; zfssync_log_file = test_files/zfssync/yoland-zfssync.log

merlin/config/test_files/eeelinux

@@ -0,0 +1,5 @@
+is_running           = false
+last_attempt_time    = 1600000000
+last_attempt_runtime = 100
+last_attempt_exit    = 1
+

merlin/config/utils.go

@@ -21,14 +21,13 @@ func touchFile(file string) {
 		f.Close()
 	} else if fi.IsDir() {
 		panic(fmt.Errorf("%s is a directory", file))
-	} else if os.Geteuid() != 1001 {
-		// UID 1001 is the hardcoded uid for mirror
-		err := os.Chown(file, 1001, os.Getegid())
-		panicIfErr(err)
-	} else if fi.Mode().Perm() != 0644 {
-		err := os.Chmod(file, 0644)
-		panicIfErr(err)
+		// } else if os.Geteuid() != 1001 {
+		// 	// mirror is UID 1001
+		// 	err := os.Chown(file, 1001, os.Getegid())
+		// 	panicIfErr(err)
 	}
+	err = os.Chmod(file, 0644)
+	panicIfErr(err)
 }
 
 func touchFiles(files ...string) {

merlin/go.mod

@@ -1,6 +1,7 @@
 module git.csclub.uwaterloo.ca/public/merlin
 
 require (
+	github.com/davecgh/go-spew v1.1.0
 	github.com/stretchr/testify v1.7.0 // indirect
 	golang.org/x/sys v0.0.0-20210915083310-ed5796bab164
 	gopkg.in/ini.v1 v1.63.2

merlin/logger/logger.go

@@ -67,7 +67,9 @@ func (logger *Logger) Debug(v ...interface{}) {
 }
 
 func (logger *Logger) Info(v ...interface{}) {
-	OutLog(append([]interface{}{"[" + logger.name + "]"}, v...))
+	// src := []interface{}{logger.name + ":"}
+	// args := append(src, v...)
+	OutLog(v...)
 	logger.Log(INFO, v...)
 }
 

merlin/merlin.go

@@ -16,6 +16,9 @@ import (
 	"git.csclub.uwaterloo.ca/public/merlin/sync"
 )
 
+// get config path from command args
+var CONFIG_PATH = "merlin-config.ini"
+
 func main() {
 	// check that merlin is run as mirror user
 	// check that mirror user has pid of 1001
@@ -40,11 +43,11 @@ func main() {
 	repoIdx := 0
 
 	loadConfig := func() {
-		config.LoadConfig(doneChan, stopChan)
+		config.LoadConfig(CONFIG_PATH, doneChan, stopChan)
 		logger.OutLog("Loaded config:\n" + fmt.Sprintf("%+v\n", config.Conf))
 
 		repoIdx = 0
-		go arthur.UnixSockListener(connChan, stopLisChan)
+		go arthur.StartListener(connChan, stopLisChan)
 	}
 	// We use a round-robin strategy. It's not the most efficient, but it's simple
 	// (read: easy to understand) and guarantees each repo will eventually get a chance to run.
@@ -66,7 +69,8 @@ func main() {
 
 	loadConfig()
 	// ensure that IsRunning is false otherwise repo will never sync
-	for _, repo := range config.RepoMap {
+	// (only on startup can we assume that repos were not previously syncing)
+	for _, repo := range config.Repos {
 		repo.State.IsRunning = false
 	}
 	runAsManyAsPossible()
@@ -82,18 +86,28 @@ runLoop:
 		case <-reloadSig:
 			stopLisChan <- struct{}{}
 			loadConfig()
+			// ensure that SyncCompleted can handle it if reloading config
+			// removes a repo that was already syncing
 
 		case done := <-doneChan:
 			sync.SyncCompleted(config.RepoMap[done.Name], done.Exit)
 			numJobsRunning--
 
 		case conn := <-connChan:
-			// TODO: may want to split this into GetCommand and something else
-			// to make it more clear tha GetAndRunCommand returns true if
-			// it starts a sync
-			if arthur.GetAndRunCommand(conn) {
-				numJobsRunning--
+			command, repoName := arthur.GetCommand(conn)
+			switch command {
+			case "status":
+				arthur.SendStatus(conn)
+			case "sync":
+				if arthur.ForceSync(conn, repoName) {
+					numJobsRunning++
+				}
+			default:
+				arthur.SendAndLog(conn, "Received unrecognized command: "+command)
 			}
+			// None of the arthur functions close the connection so you will need to
+			// close it manually for the message to be sent
+			conn.Close()
 
 		case <-time.After(1 * time.Minute):
 		}

merlin/sync/command.go

@@ -2,8 +2,7 @@ package sync
 
 import (
 	"fmt"
-	"math/rand"
-	"strconv"
+	"os"
 
 	"git.csclub.uwaterloo.ca/public/merlin/config"
 )
@@ -170,17 +169,14 @@ func cscSyncStandardIPV6(repo *config.Repo) []string {
 	return args
 }
 
-// for testing, to be removed later
 func cscSyncDummy(repo *config.Repo) []string {
-
-	sleepDur := strconv.FormatInt(rand.Int63n(10)+5, 10)
-	args := []string{"sleep", sleepDur}
+	args := []string{"sleep", "1"}
 
 	return args
 }
 
 // executes a particular sync job depending on repo.SyncType.
-func getSyncCommand(repo *config.Repo) []string {
+func getSyncCommand(repo *config.Repo) (args []string) {
 	/*
 		# scripts used by merlin.py
 		csc-sync-debian
@@ -207,40 +203,55 @@ func getSyncCommand(repo *config.Repo) []string {
 		ubuntu-releases-sync
 	*/
 	switch repo.SyncType {
-
 	case "csc-sync-apache":
-		return cscSyncApache(repo)
+		args = cscSyncApache(repo)
 	case "csc-sync-archlinux":
-		return cscSyncArchLinux(repo)
+		args = cscSyncArchLinux(repo)
 	case "csc-sync-badperms":
-		return cscSyncBadPerms(repo)
+		args = cscSyncBadPerms(repo)
 	case "csc-sync-cdimage":
-		return cscSyncCDImage(repo)
+		args = cscSyncCDImage(repo)
 	// case "csc-sync-ceph":
-	// 	return cscSyncCeph(repo)
+	// 	args = cscSyncCeph(repo)
 	case "csc-sync-chmod":
-		return cscSyncChmod(repo)
+		args = cscSyncChmod(repo)
 	case "csc-sync-debian":
-		return cscSyncDebian(repo)
+		args = cscSyncDebian(repo)
 	case "csc-sync-debian-cd":
-		return cscSyncDebianCD(repo)
+		args = cscSyncDebianCD(repo)
 	case "csc-sync-gentoo":
-		return cscSyncGentoo(repo)
+		args = cscSyncGentoo(repo)
 	// case "csc-sync-s3":
-	// 	return cscSyncS3(repo)
+	// 	args = cscSyncS3(repo)
 	case "csc-sync-ssh":
-		return cscSyncSSH(repo)
+		args = cscSyncSSH(repo)
 	case "csc-sync-standard":
-		return cscSyncStandard(repo)
+		args = cscSyncStandard(repo)
 	case "csc-sync-standard-ipv6":
-		return cscSyncStandardIPV6(repo)
+		args = cscSyncStandardIPV6(repo)
 		// case "csc-sync-wget":
-		// 	return cscSyncWget(repo)
+		// 	args = cscSyncWget(repo)
 	case "csc-sync-dummy":
 		return cscSyncDummy(repo)
 	default:
 		repo.Logger.Error("Unrecognized sync type", "'"+repo.SyncType+"'")
 	}
 
-	return []string{}
+	localDir := buildDownloadDir(repo)
+	err := os.MkdirAll(localDir, 0775)
+	if err != nil {
+		err = fmt.Errorf("could not create directory %s: %w", localDir, err)
+		// I'm not sure if logger can handle error so just use the string?
+		repo.Logger.Error(err.Error())
+		return
+	}
+
+	if repo.PasswordFile != "" {
+		filename := config.Conf.PasswordDir + "/" + repo.PasswordFile
+		args = append(args, "--password-file", filename)
+	}
+
+	args = append(args, buildRsyncHost(repo), localDir)
+
+	return
 }

merlin/sync/interface.go

@@ -52,6 +52,8 @@ func SyncCompleted(repo *config.Repo, exit int) {
 	}
 	repo.SaveState()
 	repo.Logger.Info(fmt.Sprintf("Sync "+exitStr+" after running for %d seconds, will run again in %d seconds", syncTook, nextSync))
+
+	// TODO: make it possible for this SyncCompleted to be run without zfsSync being run
 	if exit == config.SUCCESS {
 		// it is possible that the zfssync from the last repo sync is still running is that fine?
 		go zfsSync(repo)

merlin/sync/sync.go

@@ -58,6 +58,7 @@ func spawnSyncProcess(repo *config.Repo, args []string) (ch <-chan *exec.Cmd) {
 		case <-cmdDoneChan:
 			if !cmd.ProcessState.Success() {
 				repo.Logger.Warning("Process ended with status code", cmd.ProcessState.ExitCode())
+				// repo.Logger.Info(strings.Join(args, " "))
 			}
 
 		case <-repo.StopChan:
@@ -81,22 +82,7 @@ func startSync(repo *config.Repo) {
 		}
 	}()
 
-	localDir := buildDownloadDir(repo)
-	err := os.MkdirAll(localDir, 0775)
-	if err != nil {
-		err = fmt.Errorf("Could not create directory %s: %w", localDir, err)
-		// I'm not sure if logger can handle error so just use the string?
-		repo.Logger.Error(err.Error())
-		return
-	}
-
 	args := getSyncCommand(repo)
-	if repo.PasswordFile != "" {
-		filename := config.Conf.PasswordDir + "/" + repo.PasswordFile
-		args = append(args, "--password-file", filename)
-	}
-	args = append(args, buildRsyncHost(repo), localDir)
-
 	ch := spawnSyncProcess(repo, args)
 	if ch == nil {
 		// SpawnProcess will have already logged error