/* Refresh a server object */
func (sm *ServerMonitor) refresh() error {
	err := sm.Conn.Ping()
	if err != nil {
		return err
	}
	sv, err := dbhelper.GetVariables(sm.Conn)
	if err != nil {
		return err
	}
	sm.BinlogPos = sv["GTID_BINLOG_POS"]
	sm.Strict = sv["GTID_STRICT_MODE"]
	sm.LogBin = sv["LOG_BIN"]
	sm.ReadOnly = sv["READ_ONLY"]
	sm.CurrentGtid = sv["GTID_CURRENT_POS"]
	sm.SlaveGtid = sv["GTID_SLAVE_POS"]
	sid, _ := strconv.ParseUint(sv["SERVER_ID"], 10, 0)
	sm.ServerId = uint(sid)
	slaveStatus, err := dbhelper.GetSlaveStatus(sm.Conn)
	if err != nil {
		return err
	}
	sm.UsingGtid = slaveStatus.Using_Gtid
	sm.IOThread = slaveStatus.Slave_IO_Running
	sm.SQLThread = slaveStatus.Slave_SQL_Running
	sm.Delay = slaveStatus.Seconds_Behind_Master
	sm.MasterServerId = slaveStatus.Master_Server_Id
	sm.MasterHost = slaveStatus.Master_Host
	return err
}
/* Refresh a server object */
func (server *ServerMonitor) refresh() error {
	err := server.Conn.Ping()
	if err != nil {
		// we want the failed state for masters to be set by the monitor
		if server.State != stateMaster {
			server.State = stateFailed
			// remove from slave list
			server.delete(&slaves)
		}
		return err
	}
	sv, err := dbhelper.GetVariables(server.Conn)
	if err != nil {
		return err
	}
	server.PrevState = server.State
	server.BinlogPos = sv["GTID_BINLOG_POS"]
	server.Strict = sv["GTID_STRICT_MODE"]
	server.LogBin = sv["LOG_BIN"]
	server.ReadOnly = sv["READ_ONLY"]
	server.CurrentGtid = sv["GTID_CURRENT_POS"]
	server.SlaveGtid = sv["GTID_SLAVE_POS"]
	sid, _ := strconv.ParseUint(sv["SERVER_ID"], 10, 0)
	server.ServerID = uint(sid)
	slaveStatus, err := dbhelper.GetSlaveStatus(server.Conn)
	if err != nil {
		// If we reached this stage with a previously failed server, reintroduce
		// it as unconnected server.
		if server.State == stateFailed {
			server.State = stateUnconn
			if autorejoin {
				if verbose {
					logprint("INFO : Rejoining previously failed server", server.URL)
				}
				err := server.rejoin()
				if err != nil {
					logprint("ERROR: Failed to autojoin previously failed server", server.URL)
				}
			}
		}
		return err
	}
	server.UsingGtid = slaveStatus.Using_Gtid
	server.IOThread = slaveStatus.Slave_IO_Running
	server.SQLThread = slaveStatus.Slave_SQL_Running
	server.Delay = slaveStatus.Seconds_Behind_Master
	server.MasterServerID = slaveStatus.Master_Server_Id
	server.MasterHost = slaveStatus.Master_Host
	// In case of state change, reintroduce the server in the slave list
	if server.PrevState == stateFailed || server.PrevState == stateUnconn {
		server.State = stateSlave
		slaves = append(slaves, server)
	}
	return err
}
/* Returns a candidate from a list of slaves. If there's only one slave it will be the de facto candidate. */
func (master *ServerMonitor) electCandidate(l []*ServerMonitor) int {
	ll := len(l)
	if *verbose {
		logprintf("DEBUG: Processing %d candidates", ll)
	}
	seqList := make([]uint64, ll)
	i := 0
	hiseq := 0
	for _, sl := range l {
		if *failover == "" {
			if *verbose {
				logprintf("DEBUG: Checking eligibility of slave server %s", sl.URL)
			}
			if dbhelper.CheckSlavePrerequisites(sl.Conn, sl.Host) == false {
				continue
			}
			if dbhelper.CheckBinlogFilters(master.Conn, sl.Conn) == false {
				logprintf("WARN : Binlog filters differ on master and slave %s. Skipping", sl.URL)
				continue
			}
			if dbhelper.CheckReplicationFilters(master.Conn, sl.Conn) == false {
				logprintf("WARN : Replication filters differ on master and slave %s. Skipping", sl.URL)
				continue
			}
			ss, _ := dbhelper.GetSlaveStatus(sl.Conn)
			if ss.Seconds_Behind_Master.Valid == false {
				logprintf("WARN : Slave %s is stopped. Skipping", sl.URL)
				continue
			}
			if ss.Seconds_Behind_Master.Int64 > *maxDelay {
				logprintf("WARN : Slave %s has more than %d seconds of replication delay (%d). Skipping", sl.URL, *maxDelay, ss.Seconds_Behind_Master.Int64)
				continue
			}
			if *gtidCheck && dbhelper.CheckSlaveSync(sl.Conn, master.Conn) == false {
				logprintf("WARN : Slave %s not in sync. Skipping", sl.URL)
				continue
			}
		}
		/* If server is in the ignore list, do not elect it */
		if contains(ignoreList, sl.URL) {
			if *verbose {
				logprintf("DEBUG: %s is in the ignore list. Skipping", sl.URL)
			}
			continue
		}
		/* Rig the election if the examined slave is preferred candidate master */
		if sl.URL == *prefMaster {
			if *verbose {
				logprintf("DEBUG: Election rig: %s elected as preferred master", sl.URL)
			}
			return i
		}
		seqList[i] = getSeqFromGtid(dbhelper.GetVariableByName(sl.Conn, "GTID_CURRENT_POS"))
		var max uint64
		if i == 0 {
			max = seqList[0]
		} else if seqList[i] > max {
			max = seqList[i]
			hiseq = i
		}
		i++
	}
	if i > 0 {
		/* Return key of slave with the highest seqno. */
		return hiseq
	} else {
		log.Println("ERROR: No suitable candidates found.")
		return -1
	}
}
示例#4
0
func main() {
	flag.Parse()
	if *version == true {
		fmt.Println("MariaDB Replication Manager version", repmgrVersion)
	}
	// if slaves option has been supplied, split into a slice.
	if *hosts != "" {
		hostList = strings.Split(*hosts, ",")
	} else {
		log.Fatal("ERROR: No hosts list specified.")
	}
	// validate users.
	if *user == "" {
		log.Fatal("ERROR: No master user/pair specified.")
	}
	dbUser, dbPass = splitPair(*user)
	if *rpluser == "" {
		log.Fatal("ERROR: No replication user/pair specified.")
	}
	rplUser, rplPass = splitPair(*rpluser)

	// Create a connection to each host.
	hostCount := len(hostList)
	hhdls = make([]*ServerMonitor, hostCount)
	slaveCount := 0
	for k, url := range hostList {
		var err error
		hhdls[k], err = newServerMonitor(url)
		if *verbose {
			log.Printf("DEBUG: Creating new server: %v", hhdls[k].URL)
		}
		if err != nil {
			if *state == "dead" {
				log.Printf("INFO: Server %s is dead. Assuming old master.", hhdls[k].URL)
				master = hhdls[k]
				continue
			}
			log.Fatalln("ERROR: Error when establishing initial connection to host", err)
		}
		defer hhdls[k].Conn.Close()
		if *verbose {
			log.Printf("DEBUG: Checking if server %s is slave", hhdls[k].URL)
		}
		ss, err := dbhelper.GetSlaveStatus(hhdls[k].Conn)
		if ss.Master_Host != "" {
			log.Printf("INFO : Server %s is configured as a slave", hhdls[k].URL)
			slave = append(slave, hhdls[k])
			slaveCount++
		} else {
			log.Printf("INFO : Server %s is not a slave. Assuming master status.", hhdls[k].URL)
			master = hhdls[k]
		}
	}
	if (hostCount - slaveCount) == 0 {
		log.Fatalln("ERROR: Multi-master topologies are not yet supported.")
	}

	for _, sl := range slave {
		if *verbose {
			log.Printf("DEBUG: Checking if server %s is a slave of server %s", sl.Host, master.Host)
		}
		if dbhelper.IsSlaveof(sl.Conn, sl.Host, master.IP) == false {
			log.Fatalf("ERROR: Server %s is not a slave of declared master %s", master.URL, master.Host)
		}
	}

	// Check if preferred master is included in Host List
	ret := func() bool {
		for _, v := range hostList {
			if v == *prefMaster {
				return true
			}
		}
		return false
	}
	if ret() == false && *prefMaster != "" {
		log.Fatal("ERROR: Preferred master is not included in the hosts option")
	}

	// Do failover or switchover interactively, else start the interactive monitor.
	if *state == "dead" {
		master.failover()
	} else if *interactive == false {
		master.switchover()
	} else {
	MainLoop:
		err := termbox.Init()
		if err != nil {
			log.Fatalln("Termbox initialization error", err)
		}
		termboxChan := new_tb_chan()
		interval := time.Second
		ticker := time.NewTicker(interval * 3)
		var command string
		for exit == false {
			select {
			case <-ticker.C:
				drawHeader()
				master.refresh()
				master.drawMaster()
				vy = 6
				for k, _ := range slave {
					slave[k].refresh()
					slave[k].drawSlave(&vy)
				}
				drawFooter(&vy)
				termbox.Flush()
			case event := <-termboxChan:
				switch event.Type {
				case termbox.EventKey:
					if event.Key == termbox.KeyCtrlS {
						command = "switchover"
						exit = true
					}
					if event.Key == termbox.KeyCtrlF {
						command = "failover"
						exit = true
					}
					if event.Key == termbox.KeyCtrlQ {
						exit = true
					}
				}
				switch event.Ch {
				case 's':
					termbox.Sync()
				}
			}
		}
		termbox.Close()
		switch command {
		case "switchover":
			nmUrl, nsKey := master.switchover()
			if nmUrl != "" && nsKey >= 0 {
				if *verbose {
					log.Printf("DEBUG: Reinstancing new master: %s and new slave: %s [%d]", nmUrl, slave[nsKey].URL, nsKey)
				}
				master, err = newServerMonitor(nmUrl)
				slave[nsKey], err = newServerMonitor(slave[nsKey].URL)
			}
			log.Println("###### Restarting monitor console in 5 seconds. Press Ctrl-C to exit")
			time.Sleep(5 * time.Second)
			exit = false
			goto MainLoop
		case "failover":
			nmUrl, _ := master.failover()
			if nmUrl != "" {
				if *verbose {
					log.Printf("DEBUG: Reinstancing new master: %s", nmUrl)
				}
				master, err = newServerMonitor(nmUrl)
			}
			log.Println("###### Restarting monitor console in 5 seconds. Press Ctrl-C to exit")
			time.Sleep(5 * time.Second)
			exit = false
			goto MainLoop
		}
	}
}
示例#5
0
func main() {

	flag.Parse()
	if *version == true {
		fmt.Println("MariaDB Tools version 0.0.1")
		os.Exit(0)
	}
	var address string
	if *socket != "" {
		address = "unix(" + *socket + ")"
	}
	if *host != "" {
		address = "tcp(" + *host + ":" + *port + ")"
	}

	db, _ := sqlx.Open("mysql", *user+":"+*password+"@"+address+"/")
	err := db.Ping()
	if err != nil {
		log.Fatal(err)
	}

	defer db.Close()

	status = dbhelper.GetStatusAsInt(db)
	variable, _ = dbhelper.GetVariables(db)

	out, err := exec.Command("uname", "-srm").Output()
	if err != nil {
		log.Fatal(err)
	}
	hostname, _ := os.Hostname()
	fmt.Printf("### MariaDB Server report for host %s\n", hostname)
	fmt.Printf("### %-25s%s", "Kernel version", out)
	fmt.Printf("### %-25s%s\n", "System Time", time.Now().Format("2006-01-02 at 03:04 (MST)"))
	fmt.Println(common.DrawHashline("General", 60))
	var server_version string
	db.QueryRow("SELECT VERSION()").Scan(&server_version)
	pPrintStr("Version", server_version)
	now := time.Now().Unix()
	uptime := status["UPTIME"]
	start_time := time.Unix(now-uptime, 0).Local()
	pPrintStr("Started", humanize.Time(start_time))
	var count int64
	db.Get(&count, "SELECT COUNT(*) FROM information_schema.schemata")
	pPrintInt("Databases", count)
	db.Get(&count, "SELECT COUNT(*) FROM information_schema.tables")
	pPrintInt("Tables", count) /* Potentially unsafe for large systems */
	pPrintStr("Datadir", variable["DATADIR"])
	pPrintStr("Binary Log", variable["LOG_BIN"])
	if variable["LOG_BIN"] == "ON" {
		pPrintStr("Binlog writes per hour", humanize.IBytes(uint64(status["BINLOG_BYTES_WRITTEN"]/status["UPTIME"])*3600))
	}
	// Add stuff for slow logs
	slaveStatus, err := dbhelper.GetSlaveStatus(db)
	if err != nil {
		slaveIO := slaveStatus.Slave_IO_Running
		slaveSQL := slaveStatus.Slave_SQL_Running
		var slaveState string
		if slaveIO == "Yes" && slaveSQL == "Yes" {
			slaveState = "Slave configured, threads running"
		} else {
			slaveState = "Slave configured, threads stopped"
		}
		pPrintStr("Replication", slaveState)
	} else {
		pPrintStr("Replication", "Not configured")
	}

	// InnoDB
	fmt.Println(common.DrawHashline("InnoDB", 60))
	ibps := humanize.IBytes(common.StrtoUint(variable["INNODB_BUFFER_POOL_SIZE"]))
	pPrintStr("InnoDB Buffer Pool", ibps)
	ibpsPages := float64(status["INNODB_BUFFER_POOL_PAGES_TOTAL"])
	ibpsFree := float64(status["INNODB_BUFFER_POOL_PAGES_FREE"])
	ibpsUsed := common.DecimaltoPctLow(ibpsFree, ibpsPages)
	pPrintStr("InnoDB Buffer Used", strconv.Itoa(ibpsUsed)+"%")
	ibpsDirty := float64(status["INNODB_BUFFER_POOL_PAGES_DIRTY"])
	ibpsDirtyPct := common.DecimaltoPct(ibpsDirty, ibpsPages)
	pPrintStr("InnoDB Buffer Dirty", strconv.Itoa(ibpsDirtyPct)+"%")
	pPrintStr("InnoDB Log Files", string(variable["INNODB_LOG_FILES_IN_GROUP"])+" files of "+humanize.IBytes(common.StrtoUint(variable["INNODB_LOG_FILE_SIZE"])))
	pPrintStr("InnoDB log writes per hour", humanize.IBytes(uint64(status["INNODB_OS_LOG_WRITTEN"]/status["UPTIME"])*3600))
	pPrintStr("InnoDB Log Buffer", humanize.IBytes(common.StrtoUint(variable["INNODB_LOG_BUFFER_SIZE"])))
	var iftc string
	switch variable["INNODB_FLUSH_LOG_AT_TRX_COMMIT"] {
	case "0":
		iftc = "0 - Flush log and write buffer every sec"
	case "1":
		iftc = "1 - Write buffer and Flush log at each trx commit"
	case "2":
		iftc = "2 - Write buffer at each trx commit, Flush log every sec"
	}
	pPrintStr("InnoDB Flush Log", iftc)
	ifm := variable["INNODB_FLUSH_METHOD"]
	if ifm == "" {
		ifm = "fsync"
	}
	pPrintStr("InnoDB Flush Method", ifm)
	pPrintStr("InnoDB IO Capacity", variable["INNODB_IO_CAPACITY"])
	// MyISAM
	fmt.Println(common.DrawHashline("MyISAM", 60))
	kbs := humanize.IBytes(common.StrtoUint(variable["KEY_BUFFER_SIZE"]))
	pPrintStr("MyISAM Key Cache", kbs)
	kbs_free := float64(status["KEY_BLOCKS_UNUSED"])
	kbs_used := float64(status["KEY_BLOCKS_USED"])
	kbsUsedPct := int(((1 - (kbs_free / (kbs_free + kbs_used))) * 100) + 0.5)
	pPrintStr("MyISAM Cache Used", strconv.Itoa(kbsUsedPct)+"%")
	// Handlers
	pPrintInt("Open tables", status["OPEN_TABLES"])
	pPrintInt("Open files", status["OPEN_FILES"])
}