PT-2421 - Adding further data collection to pt-k8s-debug-collector

This PR introduces additional files in to the final dump archive for pgv2.
Tool logs:
- patronictl list
- pgbackrest info
Log files from folders:
- $PGDATA/log
- pgdata/pgbackrest/log

Author: BON4

src/go/pt-k8s-debug-collector/dumper/dumper.go

@@ -65,12 +65,19 @@ type Dumper struct {
 	restConfig      *rest.Config
 }
 
-// individualFile struct is used to dump the necessary files from the containers
+type toolLog struct {
+	filename string
+	args     []string
+}
+
+// individualFile struct is used to dump the necessary files from the containers,
+// or files produced by executing a tool command inside a container
 type individualFile struct {
 	resourceName  string
 	containerName string
 	filepaths     []string
 	dirpaths      map[string][]string // map[tarFolder][]dirPaths
+	toolCmds      map[string][]toolLog
 }
 
 // resourceMap struct is used to dump the resources from namespace scope or cluster scope

src/go/pt-k8s-debug-collector/dumper/individual_files.go

@@ -68,9 +68,33 @@ func (d *Dumper) getIndividualFiles(ctx context.Context, job exportJob, crType s
 				}
 			}
 		}
+
+		// Process tool command outputs (e.g. patronictl list, pgbackrest info)
+		for tarFolder, cmds := range indf.toolCmds {
+			for _, cmd := range cmds {
+				if err := d.processToolOutput(ctx, job, indf.containerName, tarFolder, cmd); err != nil {
+					log.Warnf("Skipping tool cmd %v: %v", cmd.args, err)
+				}
+			}
+		}
 	}
 }
 
+func (d *Dumper) processToolOutput(
+	ctx context.Context,
+	job exportJob,
+	container, tarFolder string, cmd toolLog,
+) error {
+	out, stderr, err := d.executeInPod(ctx, cmd.args, job.Pod, container, nil)
+	if err != nil {
+		return fmt.Errorf("exec %s: %w (stderr: %s)", cmd, err, stderr.String())
+	}
+
+	dst := d.PodIndividualFilesPath(job.Pod.Namespace, job.Pod.Name, path.Join(tarFolder, cmd.filename))
+
+	return d.archive.WriteVirtualFile(dst, out.Bytes())
+}
+
 func (d *Dumper) processSingleFile(
 	ctx context.Context,
 	job exportJob,

src/go/pt-k8s-debug-collector/dumper/resources.go

@@ -26,13 +26,28 @@ func (d *Dumper) addPg1() error {
 
 func (d *Dumper) addPg2() error {
 	dirpaths := map[string][]string{
-		"pg_log": {"$PGDATA/log"},
+		"pg_log":         {"$PGDATA/log"},
+		"pgbackrest_log": {"pgdata/pgbackrest/log"},
+	}
+
+	tools := map[string][]toolLog{
+		"": {
+			{
+				filename: "patronictl-list.log",
+				args:     []string{"patronictl", "list"},
+			},
+			{
+				filename: "pgbackrest-info.log",
+				args:     []string{"pgbackrest", "info"},
+			},
+		},
 	}
 
 	d.individualFiles = append(d.individualFiles, individualFile{
 		resourceName:  "pgv2",
 		containerName: "database",
 		dirpaths:      dirpaths,
+		toolCmds:      tools,
 	})
 	return nil
 }

src/go/pt-k8s-debug-collector/main_test.go

@@ -379,6 +379,64 @@ func (s *CollectorSuite) TestIndividualFiles() {
 				return strings.Join(result, "")
 			},
 		},
+		{
+			namespace: "pgv2",
+			// If the tool collects pgBackRest execution log files for pgv2
+			name: "pgv2_pgbackrest_log_list",
+			// tar -tf cluster-dump.tar.gz --wildcards 'cluster-dump/*/*/pgbackrest_log/*'
+			cmd:  []string{"tar", "-tf", "cluster-dump.tar.gz", "--wildcards", "cluster-dump/*/*/pgbackrest_log/*"},
+			want: []string{"db-archive-push-async.log", "db-stanza-create.log"},
+			preprocessor: func(in string) string {
+				required := map[string]struct{}{
+					"db-archive-push-async.log": {},
+					"db-stanza-create.log":      {},
+				}
+
+				files := strings.Split(in, "\n")
+				var result []string
+				for _, f := range files {
+					b := path.Base(f)
+					if _, ok := required[b]; !ok {
+						continue
+					}
+
+					if !slices.Contains(result, b) {
+						result = append(result, b)
+					}
+				}
+				slices.Sort(result)
+				return strings.Join(result, "\n")
+			},
+		},
+		{
+			namespace: "pgv2",
+			// If the tool collects tool command outputs (patronictl list, pgbackrest info) for pgv2
+			name: "pgv2_tools_log_list",
+			// tar -tf cluster-dump.tar.gz --wildcards 'cluster-dump/*/*/*'
+			cmd:  []string{"tar", "-tf", "cluster-dump.tar.gz", "--wildcards", "cluster-dump/*/*/*"},
+			want: []string{"patronictl-list.log", "pgbackrest-info.log"},
+			preprocessor: func(in string) string {
+				required := map[string]struct{}{
+					"patronictl-list.log": {},
+					"pgbackrest-info.log": {},
+				}
+
+				files := strings.Split(in, "\n")
+				var result []string
+				for _, f := range files {
+					b := path.Base(f)
+					if _, ok := required[b]; !ok {
+						continue
+					}
+
+					if !slices.Contains(result, b) {
+						result = append(result, b)
+					}
+				}
+				slices.Sort(result)
+				return strings.Join(result, "\n")
+			},
+		},
 		{
 			namespace: "pxc",
 			// If pod logs are exported as one file per container