Initial import

4 files changed, 415 insertions, 0 deletions

diff --git a/go.mod b/go.mod
new file mode 100644
index 0000000..505d991
--- /dev/null
+++ b/go.mod
@@ -0,0 +1,3 @@
+module bareos-zabbix-check
+
+go 1.13
diff --git a/main.go b/main.go
new file mode 100644
index 0000000..10a0d63
--- /dev/null
+++ b/main.go
@@ -0,0 +1,152 @@
+package main
+
+import (
+	"flag"
+	"fmt"
+	"log"
+	"os"
+	"path"
+	"time"
+)
+
+const (
+	bareosWorkDir   = "/var/lib/bareos"
+	bareosStateFile = "bareos-fd.9102.state"
+	baculaWorkDir   = "/var/lib/bacula"
+	baculaStateFile = "bacula-fd.9102.state"
+	spoolFile       = "bareos-zabbix-check.spool"
+)
+
+// We declare globally the variables that will hold the command line arguments
+var (
+	verbose   bool
+	quiet     bool
+	stateFile string
+	workDir   string
+)
+
+func main() {
+	var (
+		info           os.FileInfo
+		err            error
+		successfulJobs jobs
+		errorJobs      jobs
+		spoolJobs      jobs
+		jobName        string
+		ts             uint64
+		now            uint64
+		errorString    string
+		missingString  string
+	)
+
+	// command line arguments parsing
+	flag.BoolVar(&verbose, "v", false, "Activates verbose debugging output, defaults to false.")
+	flag.BoolVar(&quiet, "q", false, "Suppress all output, suitable to force a silent update of the spool file.")
+	flag.StringVar(&stateFile, "f", "", "Force the state file to use, defaults to "+bareosStateFile+" if it exists else "+baculaStateFile+".")
+	flag.StringVar(&workDir, "w", "", "Force the work directory to use, defaults to "+bareosWorkDir+" if it exists else "+baculaWorkDir+".")
+	flag.Parse()
+
+	// Determine the work directory to use.
+	if workDir != "" {
+		info, err = os.Stat(workDir)
+		if os.IsNotExist(err) || !info.IsDir() {
+			fmt.Printf("INFO Invalid work directory %s : it does not exist or is not a directory.\n", workDir)
+			os.Exit(0)
+		}
+	} else {
+		workDir = "/var/lib/bareos"
+		info, err = os.Stat(workDir)
+		if os.IsNotExist(err) || !info.IsDir() {
+			workDir = "/var/lib/bacula"
+			info, err := os.Stat(workDir)
+			if os.IsNotExist(err) || !info.IsDir() {
+				fmt.Println("INFO Could not find a suitable work directory. Is bareos or bacula installed?")
+				os.Exit(0)
+			}
+		}
+	}
+	workDir = path.Clean(workDir)
+	if verbose {
+		log.Println("Setting work directory to ", workDir)
+	}
+
+	// Finds the state file to parse
+	if stateFile != "" {
+		stateFile = path.Join(workDir, stateFile)
+		info, err = os.Stat(stateFile)
+		if os.IsNotExist(err) || info.IsDir() {
+			fmt.Printf("INFO The state file %s does not exist.\n", stateFile)
+			os.Exit(0)
+		}
+	} else {
+		stateFile = path.Join(workDir, bareosStateFile)
+		info, err = os.Stat(stateFile)
+		if os.IsNotExist(err) || info.IsDir() {
+			stateFile = path.Join(workDir, baculaStateFile)
+			info, err = os.Stat(stateFile)
+			if os.IsNotExist(err) || info.IsDir() {
+				fmt.Println("INFO Could not find a suitable state file. Has a job ever run?")
+				os.Exit(0)
+			}
+		}
+	}
+	if verbose {
+		log.Println("Using state file ", stateFile)
+	}
+
+	successfulJobs, errorJobs, err = parseStateFile()
+	if err != nil {
+		fmt.Print(err)
+		os.Exit(0)
+	}
+	// We will check for errors in loading the spool file only if necessary. If all jobs ran successfully without errors in the state file and we manage to write
+	// a new spool file without errors, then we will ignore any error here to avoid false positives during backup bootstrap
+	spoolJobs, err = loadSpool()
+
+	// if we have jobs in the spool we merge this list with successfull jobs from the state file
+	if err == nil {
+		for jobName, ts = range spoolJobs {
+			var (
+				current uint64
+				ok      bool
+			)
+			current, ok = successfulJobs[jobName]
+			if !ok || current < ts {
+				successfulJobs[jobName] = ts
+			}
+		}
+	}
+	// we write this new spool
+	if err2 := saveSpool(successfulJobs); err2 != nil {
+		fmt.Printf("AVERAGE: Couldn't save spool : %s\n", err2)
+		os.Exit(0)
+	}
+
+	// We build the error string listing the jobs in error
+	for jobName, ts = range errorJobs {
+		if errorString == "" {
+			errorString = fmt.Sprintf("errors: %s", jobName)
+		} else {
+			errorString = fmt.Sprintf("%s, %s", errorString, jobName)
+		}
+	}
+	now = uint64(time.Now().Unix())
+	// Next we check if all jobs ran recently and build the missing string
+	for jobName, ts = range successfulJobs {
+		if ts < now-24*3600 {
+			if missingString == "" {
+				missingString = fmt.Sprintf("missing: %s", jobName)
+			} else {
+				missingString = fmt.Sprintf("%s, %s", missingString, jobName)
+			}
+		}
+	}
+	if errorString != "" || missingString != "" {
+		fmt.Printf("AVERAGE: %s %s", errorString, missingString)
+		if err != nil {
+			fmt.Printf(" additionnal errors: %s", err)
+		}
+	} else {
+		fmt.Printf("OK")
+	}
+}
diff --git a/spool.go b/spool.go
new file mode 100644
index 0000000..e3d00d9
--- /dev/null
+++ b/spool.go
@@ -0,0 +1,72 @@
+package main
+
+import (
+	"encoding/csv"
+	"fmt"
+	"log"
+	"os"
+	"path"
+	"strconv"
+)
+
+// jobs is a map that matches a job name string to its last successfull run timestamp
+type jobs map[string]uint64
+
+func loadSpool() (entries jobs, err error) {
+	var (
+		file  *os.File
+		lines [][]string
+	)
+	// We read the spool
+	file, err = os.Open(path.Join(workDir, spoolFile))
+	if err != nil {
+		return nil, fmt.Errorf("INFO Couldn't open spool file: %s", err)
+	}
+	defer file.Close()
+	lines, err = csv.NewReader(file).ReadAll()
+	if err != nil {
+		return nil, fmt.Errorf("INFO Corrupted spool file : %s", err)
+	}
+	if verbose {
+		log.Printf("Spool file content : %v\n", lines)
+	}
+
+	entries = make(map[string]uint64)
+	for _, line := range lines {
+		var (
+			i int
+		)
+		i, err = strconv.Atoi(line[1])
+		if err != nil {
+			return nil, fmt.Errorf("INFO Corrupted spool file : couldn't parse timestamp entry")
+		}
+		entries[line[0]] = uint64(i)
+	}
+	return
+}
+
+func saveSpool(entries jobs) (err error) {
+	var (
+		file    *os.File
+		lines   [][]string
+		jobName string
+		ts      uint64
+		i       int
+	)
+	file, err = os.Create(path.Join(workDir, spoolFile))
+	if err != nil {
+		return fmt.Errorf("INFO Couldn't open spool file for writing : %s", err)
+	}
+	defer file.Close()
+
+	lines = make([][]string, len(entries))
+	i = 0
+	for jobName, ts = range entries {
+		lines[i] = make([]string, 2)
+		lines[i][0] = jobName
+		lines[i][1] = fmt.Sprintf("%d", ts)
+		i++
+	}
+	err = csv.NewWriter(file).WriteAll(lines)
+	return
+}
diff --git a/state.go b/state.go
new file mode 100644
index 0000000..2cfb5df
--- /dev/null
+++ b/state.go
@@ -0,0 +1,188 @@
+package main
+
+import (
+	"bytes"
+	"encoding/binary"
+	"fmt"
+	"log"
+	"os"
+	"regexp"
+)
+
+// stateFileHeader : A structure to hold the header of the state file. It is statically aligned for amd64 architecture
+type stateFileHeader struct {
+	ID                        [14]byte
+	_                         int16
+	Version                   int32
+	_                         int32
+	LastJobsAddr              uint64
+	EndOfRecentJobResultsList uint64
+	Reserved                  [19]uint64
+}
+
+func (sfh stateFileHeader) String() string {
+	return fmt.Sprintf("ID: \"%s\", Version: %d, LastJobsAddr: %d, EndOfRecentJobResultsList: %d", sfh.ID[:len(sfh.ID)-2], sfh.Version, sfh.EndOfRecentJobResultsList, sfh.Reserved)
+}
+
+// jobEntry : A structure to hold a job result from the state file
+type jobEntry struct {
+	Pad            [16]byte
+	Errors         int32
+	JobType        int32
+	JobStatus      int32
+	JobLevel       int32
+	JobID          uint32
+	VolSessionID   uint32
+	VolSessionTime uint32
+	JobFiles       uint32
+	JobBytes       uint64
+	StartTime      uint64
+	EndTime        uint64
+	Job            [maxNameLength]byte
+}
+
+func (je jobEntry) String() string {
+	var matches = jobNameRegex.FindSubmatchIndex(je.Job[:])
+	var jobNameLen int
+	if len(matches) >= 4 {
+		jobNameLen = matches[3]
+	}
+	return fmt.Sprintf("Errors: %d, JobType: %c, JobStatus: %c, JobLevel: %c, JobID: %d, VolSessionID: %d, VolSessionTime: %d, JobFiles: %d, JobBytes: %d, StartTime: %d, EndTime: %d, Job: %s",
+		je.Errors, je.JobType, je.JobStatus, je.JobLevel, je.JobID, je.VolSessionID, je.VolSessionTime, je.JobFiles, je.JobBytes, je.StartTime, je.EndTime, je.Job[:jobNameLen])
+}
+
+const (
+	// maxNameLength : the maximum length of a string, hard coded in bareos
+	maxNameLength = 128
+	// stateFileHeaderLength : the length of the state file header struct
+	stateFileHeaderLength = 14 + 2 + 4 + 4 + 8 + 8 + 19*8
+	// jobResultLength : the length of the job result struct
+	jobResultLength = 16 + 4 + 4 + 4 + 4 + 4 + 4 + 4 + 4 + 8 + 8 + 8 + maxNameLength
+)
+
+var jobNameRegex = regexp.MustCompilePOSIX(`^([-A-Za-z0-9_]+)\.[0-9]{4}-[0-9]{2}-[0-9]{2}.*`)
+
+// readNextBytes : Reads the next "number" bytes from a "file", returns the number of bytes actually read as well as the bytes read
+func readNextBytes(file *os.File, number int) (int, []byte) {
+	var bytes = make([]byte, number)
+
+	var n, err = file.Read(bytes)
+	if err != nil {
+		fmt.Printf("INFO Corrupted state file : file.Read failed in %s : %s\n", stateFile, err)
+	}
+
+	return n, bytes
+}
+
+func parseStateFile() (successfulJobs jobs, errorJobs jobs, err error) {
+	var (
+		n               int
+		stateFileHandle *os.File
+		data            []byte
+		buffer          *bytes.Buffer
+		numberOfJobs    uint32
+		matches         []int
+	)
+	// Open the state file
+	stateFileHandle, err = os.Open(stateFile)
+	if err != nil {
+		return nil, nil, fmt.Errorf("INFO Couldn't open state file : %s", err)
+	}
+	defer stateFileHandle.Close()
+
+	// Parsing the state file header
+	var header stateFileHeader
+	n, data = readNextBytes(stateFileHandle, stateFileHeaderLength)
+	if n != stateFileHeaderLength {
+		return nil, nil, fmt.Errorf("INFO Corrupted state file : invalid header length in %s", stateFile)
+	}
+	buffer = bytes.NewBuffer(data)
+	err = binary.Read(buffer, binary.LittleEndian, &header)
+	if err != nil {
+		return nil, nil, fmt.Errorf("INFO Corrupted state file : binary.Read failed on header in %s : %s", stateFile, err)
+	}
+	if verbose {
+		log.Printf("Parsed header: %+s\n", header)
+	}
+	if id := string(header.ID[:len(header.ID)-1]); id != "Bareos State\n" && id != "Bacula State\n" {
+		return nil, nil, fmt.Errorf("INFO Corrupted state file : Not a bareos or bacula state file %s", stateFile)
+	}
+	if header.Version != 4 {
+		return nil, nil, fmt.Errorf("INFO Invalid state file : This script only supports bareos state file version 4, got %d", header.Version)
+	}
+	if header.LastJobsAddr == 0 {
+		return nil, nil, fmt.Errorf("INFO No jobs exist in the state file")
+	}
+
+	// We seek to the jobs position in the state file
+	stateFileHandle.Seek(int64(header.LastJobsAddr), 0)
+
+	// We read how many jobs there are in the state file
+	n, data = readNextBytes(stateFileHandle, 4)
+	if n != 4 {
+		return nil, nil, fmt.Errorf("INFO Corrupted state file : invalid numberOfJobs read length in %s", stateFile)
+	}
+	buffer = bytes.NewBuffer(data)
+	err = binary.Read(buffer, binary.LittleEndian, &numberOfJobs)
+	if err != nil {
+		return nil, nil, fmt.Errorf("INFO Corrupted state file : binary.Read failed on numberOfJobs in %s : %s", stateFile, err)
+	}
+	if verbose {
+		log.Printf("%d jobs found in state file\n", numberOfJobs)
+	}
+
+	// We parse the job entries
+	successfulJobs = make(map[string]uint64)
+	errorJobs = make(map[string]uint64)
+	for ; numberOfJobs > 0; numberOfJobs-- {
+		var (
+			jobResult jobEntry
+			jobName   string
+		)
+		n, data = readNextBytes(stateFileHandle, jobResultLength)
+		if n != jobResultLength {
+			return nil, nil, fmt.Errorf("INFO Corrupted state file : invalid job entry in %s", stateFile)
+		}
+		buffer = bytes.NewBuffer(data)
+		err = binary.Read(buffer, binary.LittleEndian, &jobResult)
+		if err != nil {
+			return nil, nil, fmt.Errorf("INFO Corrupted state file : binary.Read failed on job entry in %s : %s", stateFile, err)
+		}
+		matches = jobNameRegex.FindSubmatchIndex(jobResult.Job[:])
+		if len(matches) >= 4 {
+			jobName = string(jobResult.Job[:matches[3]])
+		} else {
+			return nil, nil, fmt.Errorf("INFO Couldn't parse job name, this shouldn't happen : %s", jobResult.Job[:])
+		}
+		if verbose {
+			log.Printf("Parsed job entry: %s\n", jobResult)
+		}
+		// If the job is of type backup (B == ascii 66)
+		if jobResult.JobType == 66 {
+			var (
+				successExists  bool
+				errorExists    bool
+				currentSuccess uint64
+				currentError   uint64
+			)
+			currentSuccess, successExists = successfulJobs[jobName]
+			currentError, errorExists = errorJobs[jobName]
+			// If the job is of status success (T == ascii 84)
+			if jobResult.JobStatus == 84 {
+				// if there is an older entry in errorJobs we delete it
+				if errorExists && jobResult.StartTime > currentError {
+					delete(errorJobs, jobName)
+				}
+				// if there are no entries more recent in successfulJobs we add this one
+				if !successExists || successExists && jobResult.StartTime > currentSuccess {
+					successfulJobs[jobName] = jobResult.StartTime
+				}
+			} else {
+				if !errorExists || jobResult.StartTime > currentError {
+					errorJobs[jobName] = jobResult.StartTime
+				}
+			}
+		}
+	}
+	return
+}