Add Diagnostics (#203)

* Add container suplimentary groups support

* Add diagnostic gathering

* Fix incorrect userid group searching

* one last tiny fix

* one last tiny fix

Author: parrobe

cmd/runmqserver/logging.go

@@ -21,16 +21,20 @@ import (
 	"fmt"
 	"io/ioutil"
 	"os"
+	"os/exec"
 	"path/filepath"
 	"sync"
 
+	"github.com/ibm-messaging/mq-container/internal/command"
 	"github.com/ibm-messaging/mq-container/internal/logger"
 	"github.com/ibm-messaging/mq-container/internal/mqini"
 )
 
 // var debug = false
 var log *logger.Logger
 
+var collectDiagOnFail bool = false
+
 func logTerminationf(format string, args ...interface{}) {
 	logTermination(fmt.Sprintf(format, args))
 }
@@ -45,6 +49,10 @@ func logTermination(args ...interface{}) {
 		log.Debug(err)
 	}
 	log.Error(msg)
+
+	if collectDiagOnFail {
+		logDiagnostics()
+	}
 }
 
 func getLogFormat() string {
@@ -111,3 +119,27 @@ func configureLogger(name string) (mirrorFunc, error) {
 		return nil, fmt.Errorf("invalid value for LOG_FORMAT: %v", f)
 	}
 }
+
+func logDiagnostics() {
+	log.Debug("--- Start Diagnostics ---")
+
+	// show the directory ownership/permissions
+	out, _, _ := command.Run("ls", "-l", "/mnt/")
+	log.Debugf("/mnt/:\n%s", out)
+	out, _, _ = command.Run("ls", "-l", "/mnt/mqm")
+	log.Debugf("/mnt/mqm:\n%s", out)
+	out, _, _ = command.Run("ls", "-l", "/mnt/mqm/data")
+	log.Debugf("/mnt/mqm/data:\n%s", out)
+	out, _, _ = command.Run("ls", "-l", "/var/mqm")
+	log.Debugf("/var/mqm:\n%s", out)
+	out, _, _ = command.Run("ls", "-l", "/var/mqm/errors")
+	log.Debugf("/var/mqm/errors:\n%s", out)
+
+	// Print out summary of any FDCs
+	cmd := exec.Command("/opt/mqm/bin/ffstsummary")
+	cmd.Dir = "/var/mqm/errors"
+	outB, _ := cmd.CombinedOutput()
+	log.Debugf("ffstsummary:\n%s", string(outB))
+
+	log.Debug("---  End Diagnostics  ---")
+}

cmd/runmqserver/main.go

@@ -58,14 +58,16 @@ func doMain() error {
 
 	// Start signal handler
 	signalControl := signalHandler(name)
+	// Enable diagnostic collecting on failure
+	collectDiagOnFail = true
 
-	err = logConfig()
+	err = verifyCurrentUser()
 	if err != nil {
 		logTermination(err)
 		return err
 	}
 
-	err = verifyCurrentUser()
+	err = logConfig()
 	if err != nil {
 		logTermination(err)
 		return err

cmd/runmqserver/mirror.go

@@ -44,7 +44,6 @@ func waitForFile(ctx context.Context, path string) (os.FileInfo, error) {
 					return nil, fmt.Errorf("mirror: unable to get info on file %v", path)
 				}
 			}
-			log.Debugf("File exists: %v, %v", path, fi.Size())
 			return fi, nil
 		}
 	}
@@ -121,6 +120,7 @@ func mirrorLog(ctx context.Context, wg *sync.WaitGroup, path string, fromStart b
 			if fi == nil {
 				return
 			}
+			log.Debugf("File exists: %v, %v", path, fi.Size())
 			f, err = os.OpenFile(path, os.O_RDONLY, 0)
 			if err != nil {
 				log.Error(err)

cmd/runmqserver/user.go

@@ -23,7 +23,7 @@ import (
 	"github.com/ibm-messaging/mq-container/internal/command"
 )
 
-const groupName string = "suplgroup"
+const groupName string = "supplgrp"
 
 func verifyCurrentUser() error {
 	log.Debug("Verifying current user information")
@@ -36,9 +36,9 @@ func verifyCurrentUser() error {
 		// Not supported yet
 		return fmt.Errorf("Container is running as mqm user which is not supported. Please run this container as root")
 	} else if curUser.Username == "root" {
-		// We're running as root so need to check for suplimentary groups.
-		// We can't use the golang User.GroupIDs as it doesn't seem to detect container supplimentary groups..
-		groups, err := getCurrentGroups()
+		// We're running as root so need to check for supplementary groups.
+		// We can't use the golang User.GroupIDs as it doesn't seem to detect container supplementary groups..
+		groups, err := getCurrentUserGroups()
 		for _, e := range groups {
 			_, _, testGroup := command.Run("getent", "group", e)
 			if testGroup != nil {
@@ -64,9 +64,9 @@ func verifyCurrentUser() error {
 }
 
 func logUser() {
-	u, err := user.Current()
-	if err == nil {
-		g, err := getCurrentGroups()
+	u, usererr := user.Current()
+	if usererr == nil {
+		g, err := getCurrentUserGroups()
 		if err != nil && len(g) == 0 {
 			log.Printf("Running as user ID %v (%v) with primary group %v", u.Uid, u.Name, u.Gid)
 		} else {
@@ -77,12 +77,30 @@ func logUser() {
 					g = append(g[:i], g[i+1:]...)
 				}
 			}
-			log.Printf("Running as user ID %v (%v) with primary group %v, and supplemental groups %v", u.Uid, u.Name, u.Gid, strings.Join(g, ","))
+			log.Printf("Running as user ID %v (%v) with primary group %v, and supplementary groups %v", u.Uid, u.Name, u.Gid, strings.Join(g, ","))
+		}
+	}
+
+	if usererr == nil && u.Username != "mqm" {
+		mqm, err := user.Lookup("mqm")
+		// Need to print out mqm user details as well.
+		g, err := getUserGroups(mqm)
+		if err != nil && len(g) == 0 {
+			log.Printf("MQM user ID %v (%v) has primary group %v", mqm.Uid, "mqm", mqm.Gid)
+		} else {
+			// Look for the primary group in the list of group IDs
+			for i, v := range g {
+				if v == mqm.Gid {
+					// Remove the element from the slice
+					g = append(g[:i], g[i+1:]...)
+				}
+			}
+			log.Printf("MQM user ID %v (%v) has primary group %v, and supplementary groups %v", mqm.Uid, "mqm", mqm.Gid, strings.Join(g, ","))
 		}
 	}
 }
 
-func getCurrentGroups() ([]string, error) {
+func getCurrentUserGroups() ([]string, error) {
 	var nilArray []string
 	out, _, err := command.Run("id", "--groups")
 	if err != nil {
@@ -99,3 +117,21 @@ func getCurrentGroups() ([]string, error) {
 	groups := strings.Split(out, " ")
 	return groups, nil
 }
+
+func getUserGroups(usr *user.User) ([]string, error) {
+	var nilArray []string
+	out, _, err := command.Run("id", "--groups", usr.Uid)
+	if err != nil {
+		log.Debugf("Unable to get user %s groups", usr.Uid)
+		return nilArray, err
+	}
+
+	out = strings.TrimSpace(out)
+	if out == "" {
+		// we don't have any groups?
+		return nilArray, fmt.Errorf("Unable to determine groups for user %s", usr.Uid)
+	}
+
+	groups := strings.Split(out, " ")
+	return groups, nil
+}