feat(support): collect system info bundle to assist support troublesh…

…ooting [r8s-157] (#154)

docker/snapshot.go

@@ -2,20 +2,12 @@ package docker
 
 import (
 	"context"
-	"strings"
-	"time"
 
 	portainer "github.com/portainer/portainer/api"
-
-	"github.com/docker/docker/api/types"
-	"github.com/docker/docker/api/types/container"
-	"github.com/docker/docker/api/types/image"
-	"github.com/docker/docker/api/types/volume"
-	"github.com/docker/docker/client"
-	"github.com/rs/zerolog/log"
+	"github.com/portainer/portainer/pkg/snapshot"
 )
 
-func CreateSnapshot() (*portainer.DockerSnapshot, error) {
+func CreateSnapshot(edgeKey string) (*portainer.DockerSnapshot, error) {
 	cli, err := NewClient()
 	if err != nil {
 		return nil, err
@@ -27,213 +19,16 @@ func CreateSnapshot() (*portainer.DockerSnapshot, error) {
 		return nil, err
 	}
 
-	snapshot := &portainer.DockerSnapshot{
-		StackCount: 0,
-	}
-
-	err = snapshotInfo(snapshot, cli)
-	if err != nil {
-		log.Warn().Err(err).Msg("unable to snapshot engine information")
-	}
-
-	if snapshot.Swarm {
-		err = snapshotSwarmServices(snapshot, cli)
-		if err != nil {
-			log.Warn().Err(err).Msg("unable to snapshot Swarm services")
-		}
-
-		err = snapshotNodes(snapshot, cli)
-		if err != nil {
-			log.Warn().Err(err).Msg("unable to snapshot Swarm nodes")
-		}
-	}
-
-	err = snapshotContainers(snapshot, cli)
-	if err != nil {
-		log.Warn().Err(err).Msg("unable to snapshot containers")
-	}
-
-	err = snapshotImages(snapshot, cli)
-	if err != nil {
-		log.Warn().Err(err).Msg("unable to snapshot images")
-	}
-
-	err = snapshotVolumes(snapshot, cli)
-	if err != nil {
-		log.Warn().Err(err).Msg("unable to snapshot volumes")
-	}
-
-	err = snapshotNetworks(snapshot, cli)
-	if err != nil {
-		log.Warn().Err(err).Msg("unable to snapshot networks")
-	}
-
-	err = snapshotVersion(snapshot, cli)
-	if err != nil {
-		log.Warn().Err(err).Msg("unable to snapshot engine version")
-	}
-
-	snapshot.Time = time.Now().Unix()
-
-	return snapshot, nil
-}
-
-func snapshotInfo(snapshot *portainer.DockerSnapshot, cli *client.Client) error {
-	info, err := cli.Info(context.Background())
-	if err != nil {
-		return err
-	}
-
-	snapshot.Swarm = info.Swarm.ControlAvailable
-	snapshot.DockerVersion = info.ServerVersion
-	snapshot.TotalCPU = info.NCPU
-	snapshot.TotalMemory = info.MemTotal
-	snapshot.SnapshotRaw.Info = info
-
-	return nil
-}
-
-func snapshotNodes(snapshot *portainer.DockerSnapshot, cli *client.Client) error {
-	nodes, err := cli.NodeList(context.Background(), types.NodeListOptions{})
-	if err != nil {
-		return err
-	}
-	var nanoCpus int64
-	var totalMem int64
-	for _, node := range nodes {
-		nanoCpus += node.Description.Resources.NanoCPUs
-		totalMem += node.Description.Resources.MemoryBytes
-	}
-	snapshot.TotalCPU = int(nanoCpus / 1e9)
-	snapshot.TotalMemory = totalMem
-	snapshot.NodeCount = len(nodes)
-
-	return nil
-}
-
-func snapshotSwarmServices(snapshot *portainer.DockerSnapshot, cli *client.Client) error {
-	stacks := make(map[string]struct{})
-
-	services, err := cli.ServiceList(context.Background(), types.ServiceListOptions{})
-	if err != nil {
-		return err
-	}
-
-	for _, service := range services {
-		for k, v := range service.Spec.Labels {
-			if k == "com.docker.stack.namespace" {
-				stacks[v] = struct{}{}
-			}
-		}
-	}
-
-	snapshot.ServiceCount = len(services)
-	snapshot.StackCount += len(stacks)
-
-	return nil
-}
-
-func snapshotContainers(snapshot *portainer.DockerSnapshot, cli *client.Client) error {
-	rawContainers, err := cli.ContainerList(context.Background(), container.ListOptions{All: true})
+	dockerSnapshot, err := snapshot.CreateDockerSnapshot(cli)
 	if err != nil {
-		return err
-	}
-
-	runningContainers := 0
-	stoppedContainers := 0
-	healthyContainers := 0
-	unhealthyContainers := 0
-	stacks := make(map[string]struct{})
-
-	containers := make([]portainer.DockerContainerSnapshot, 0)
-
-	for _, container := range rawContainers {
-		response, err := cli.ContainerInspect(context.Background(), container.ID)
-		if err != nil {
-			log.Warn().Err(err).Msg("failed to retrieve env for container " + container.ID + ". Skipping.")
-			containers = append(containers, portainer.DockerContainerSnapshot{Container: container})
-
-			continue
-		}
-
-		containers = append(containers, portainer.DockerContainerSnapshot{
-			Container: container,
-			Env:       response.Config.Env,
-		})
-	}
-
-	for _, container := range containers {
-		if container.State == "exited" {
-			stoppedContainers++
-		} else if container.State == "running" {
-			runningContainers++
-		}
-
-		if strings.Contains(container.Status, "(healthy)") {
-			healthyContainers++
-		} else if strings.Contains(container.Status, "(unhealthy)") {
-			unhealthyContainers++
-		}
-
-		for k, v := range container.Labels {
-			if k == "com.docker.compose.project" {
-				stacks[v] = struct{}{}
-			}
-		}
-	}
-
-	snapshot.RunningContainerCount = runningContainers
-	snapshot.StoppedContainerCount = stoppedContainers
-	snapshot.HealthyContainerCount = healthyContainers
-	snapshot.UnhealthyContainerCount = unhealthyContainers
-	snapshot.StackCount += len(stacks)
-	snapshot.SnapshotRaw.Containers = containers
-
-	return nil
-}
-
-func snapshotImages(snapshot *portainer.DockerSnapshot, cli *client.Client) error {
-	images, err := cli.ImageList(context.Background(), image.ListOptions{})
-	if err != nil {
-		return err
-	}
-
-	snapshot.ImageCount = len(images)
-	snapshot.SnapshotRaw.Images = images
-
-	return nil
-}
-
-func snapshotVolumes(snapshot *portainer.DockerSnapshot, cli *client.Client) error {
-	volumes, err := cli.VolumeList(context.Background(), volume.ListOptions{})
-	if err != nil {
-		return err
-	}
-
-	snapshot.VolumeCount = len(volumes.Volumes)
-	snapshot.SnapshotRaw.Volumes = volumes
-
-	return nil
-}
-
-func snapshotNetworks(snapshot *portainer.DockerSnapshot, cli *client.Client) error {
-	networks, err := cli.NetworkList(context.Background(), types.NetworkListOptions{})
-	if err != nil {
-		return err
+		return nil, err
 	}
 
-	snapshot.SnapshotRaw.Networks = networks
-
-	return nil
-}
-
-func snapshotVersion(snapshot *portainer.DockerSnapshot, cli *client.Client) error {
-	version, err := cli.ServerVersion(context.Background())
+	diagnosticsData, err := snapshot.DockerSnapshotDiagnostics(cli, edgeKey)
 	if err != nil {
-		return err
+		return nil, err
 	}
+	dockerSnapshot.DiagnosticsData = diagnosticsData
 
-	snapshot.SnapshotRaw.Version = version
-
-	return nil
+	return dockerSnapshot, nil
 }

edge/client/interface.go

@@ -89,9 +89,9 @@ type setEndpointIDFn func(portainer.EndpointID)
 type getEndpointIDFn func() portainer.EndpointID
 
 // NewPortainerClient returns a pointer to a new PortainerClient instance
-func NewPortainerClient(serverAddress string, setEIDFn setEndpointIDFn, getEIDFn getEndpointIDFn, edgeID string, edgeAsyncMode bool, agentPlatform agent.ContainerPlatform, metaFields agent.EdgeMetaFields, httpClient *edgeHTTPClient) PortainerClient {
+func NewPortainerClient(serverAddress string, setEIDFn setEndpointIDFn, getEIDFn getEndpointIDFn, edgeID string, edgeKey string, edgeAsyncMode bool, agentPlatform agent.ContainerPlatform, metaFields agent.EdgeMetaFields, httpClient *edgeHTTPClient) PortainerClient {
 	if edgeAsyncMode {
-		return NewPortainerAsyncClient(serverAddress, setEIDFn, getEIDFn, edgeID, agentPlatform, metaFields, httpClient)
+		return NewPortainerAsyncClient(serverAddress, setEIDFn, getEIDFn, edgeID, edgeKey, agentPlatform, metaFields, httpClient)
 	}
 
 	return NewPortainerEdgeClient(serverAddress, setEIDFn, getEIDFn, edgeID, agentPlatform, metaFields, httpClient)

edge/client/portainer_edge_async_client.go

@@ -31,6 +31,7 @@ type PortainerAsyncClient struct {
 	setEndpointIDFn         setEndpointIDFn
 	getEndpointIDFn         getEndpointIDFn
 	edgeID                  string
+	edgeKey                 string
 	agentPlatformIdentifier agent.ContainerPlatform
 	commandTimestamp        *time.Time
 	metaFields              agent.EdgeMetaFields
@@ -45,13 +46,14 @@ type PortainerAsyncClient struct {
 }
 
 // NewPortainerAsyncClient returns a pointer to a new PortainerAsyncClient instance
-func NewPortainerAsyncClient(serverAddress string, setEIDFn setEndpointIDFn, getEIDFn getEndpointIDFn, edgeID string, containerPlatform agent.ContainerPlatform, metaFields agent.EdgeMetaFields, httpClient *edgeHTTPClient) *PortainerAsyncClient {
+func NewPortainerAsyncClient(serverAddress string, setEIDFn setEndpointIDFn, getEIDFn getEndpointIDFn, edgeID string, edgeKey string, containerPlatform agent.ContainerPlatform, metaFields agent.EdgeMetaFields, httpClient *edgeHTTPClient) *PortainerAsyncClient {
 	initialCommandTimestamp := time.Date(2000, 1, 1, 0, 0, 0, 0, time.UTC)
 	return &PortainerAsyncClient{
 		serverAddress:           serverAddress,
 		setEndpointIDFn:         setEIDFn,
 		getEndpointIDFn:         getEIDFn,
 		edgeID:                  edgeID,
+		edgeKey:                 edgeKey,
 		httpClient:              httpClient,
 		agentPlatformIdentifier: containerPlatform,
 		commandTimestamp:        &initialCommandTimestamp,
@@ -184,10 +186,9 @@ func (client *PortainerAsyncClient) GetEnvironmentStatus(flags ...string) (*Poll
 	var currentSnapshot snapshot
 	if doSnapshot {
 		payload.Snapshot = &snapshot{}
-
 		switch client.agentPlatformIdentifier {
 		case agent.PlatformDocker:
-			dockerSnapshot, err := docker.CreateSnapshot()
+			dockerSnapshot, err := docker.CreateSnapshot(client.edgeKey)
 			if err != nil {
 				log.Warn().Err(err).Msg("could not create the Docker snapshot")
 			}
@@ -259,7 +260,7 @@ func (client *PortainerAsyncClient) GetEnvironmentStatus(flags ...string) (*Poll
 			}
 
 		case agent.PlatformKubernetes:
-			kubeSnapshot, err := kubernetes.CreateSnapshot()
+			kubeSnapshot, err := kubernetes.CreateSnapshot(client.edgeKey)
 			if err != nil {
 				log.Warn().Err(err).Msg("could not create the Kubernetes snapshot")
 			}

edge/edge.go

@@ -104,6 +104,7 @@ func (manager *Manager) Start() error {
 		manager.SetEndpointID,
 		manager.GetEndpointID,
 		manager.agentOptions.EdgeID,
+		manager.agentOptions.EdgeKey,
 		manager.agentOptions.EdgeAsyncMode,
 		agentPlatform,
 		manager.agentOptions.EdgeMetaFields,

edge/scheduler/logs_test.go

@@ -14,6 +14,7 @@ func TestDataRace(t *testing.T) {
 		func(portainer.EndpointID) {},
 		func() portainer.EndpointID { return 1 },
 		"edgeID",
+		"edgeKey",
 		false,
 		agent.PlatformDocker,
 		agent.EdgeMetaFields{},

go.mod

@@ -28,7 +28,7 @@ require (
 	github.com/pkg/errors v0.9.1
 	github.com/portainer/portainer v0.6.1-0.20241105012540-458d722d4754
 	github.com/rs/zerolog v1.29.0
-	github.com/stretchr/testify v1.9.0
+	github.com/stretchr/testify v1.10.0
 	github.com/wI2L/jsondiff v0.2.0
 	go.uber.org/mock v0.4.0
 	gopkg.in/alecthomas/kingpin.v2 v2.2.6
@@ -160,6 +160,7 @@ require (
 	github.com/munnerz/goautoneg v0.0.0-20191010083416-a7dc8b61c822 // indirect
 	github.com/mxk/go-flowrate v0.0.0-20140419014527-cca7078d478f // indirect
 	github.com/opencontainers/go-digest v1.0.0 // indirect
+	github.com/opentracing/opentracing-go v1.2.0 // indirect
 	github.com/pelletier/go-toml v1.9.5 // indirect
 	github.com/pmezard/go-difflib v1.0.0 // indirect
 	github.com/prometheus/client_golang v1.17.0 // indirect
@@ -168,6 +169,7 @@ require (
 	github.com/prometheus/procfs v0.12.0 // indirect
 	github.com/r3labs/sse v0.0.0-20210224172625-26fe804710bc // indirect
 	github.com/rivo/uniseg v0.4.4 // indirect
+	github.com/rogpeppe/go-internal v1.11.0 // indirect
 	github.com/sean-/seed v0.0.0-20170313163322-e2103e2c3529 // indirect
 	github.com/secure-systems-lab/go-securesystemslib v0.8.0 // indirect
 	github.com/segmentio/asm v1.1.3 // indirect

go.sum

@@ -506,8 +506,9 @@ github.com/opencontainers/runtime-spec v1.2.0 h1:z97+pHb3uELt/yiAWD691HNHQIF07bE
 github.com/opencontainers/runtime-spec v1.2.0/go.mod h1:jwyrGlmzljRJv/Fgzds9SsS/C5hL+LL3ko9hs6T5lQ0=
 github.com/opencontainers/selinux v1.11.0 h1:+5Zbo97w3Lbmb3PeqQtpmTkMwsW5nRI3YaLpt7tQ7oU=
 github.com/opencontainers/selinux v1.11.0/go.mod h1:E5dMC3VPuVvVHDYmi78qvhJp8+M586T4DlDRYpFkyec=
-github.com/opentracing/opentracing-go v1.1.0 h1:pWlfV3Bxv7k65HYwkikxat0+s3pV4bsqf19k25Ur8rU=
 github.com/opentracing/opentracing-go v1.1.0/go.mod h1:UkNAQd3GIcIGf0SeVgPpRdFStlNbqXla1AfSYxPUl2o=
+github.com/opentracing/opentracing-go v1.2.0 h1:uEJPy/1a5RIPAJ0Ov+OIO8OxWu77jEv+1B0VhjKrZUs=
+github.com/opentracing/opentracing-go v1.2.0/go.mod h1:GxEUsuufX4nBwe+T+Wl9TAgYrxe9dPLANfrWvHYVTgc=
 github.com/pascaldekloe/goe v0.0.0-20180627143212-57f6aae5913c/go.mod h1:lzWF7FIEvWOWxwDKqyGYQf6ZUaNfKdP144TG7ZOy1lc=
 github.com/pascaldekloe/goe v0.1.0 h1:cBOtyMzM9HTpWjXfbbunk26uA6nG3a8n06Wieeh0MwY=
 github.com/pascaldekloe/goe v0.1.0/go.mod h1:lzWF7FIEvWOWxwDKqyGYQf6ZUaNfKdP144TG7ZOy1lc=
@@ -552,8 +553,8 @@ github.com/r3labs/sse v0.0.0-20210224172625-26fe804710bc/go.mod h1:S8xSOnV3CgpNr
 github.com/rivo/uniseg v0.2.0/go.mod h1:J6wj4VEh+S6ZtnVlnTBMWIodfgj8LQOQFoIToxlJtxc=
 github.com/rivo/uniseg v0.4.4 h1:8TfxU8dW6PdqD27gjM8MVNuicgxIjxpm4K7x4jp8sis=
 github.com/rivo/uniseg v0.4.4/go.mod h1:FN3SvrM+Zdj16jyLfmOkMNblXMcoc8DfTHruCPUcx88=
-github.com/rogpeppe/go-internal v1.10.0 h1:TMyTOH3F/DB16zRVcYyreMH6GnZZrwQVAoYjRBZyWFQ=
-github.com/rogpeppe/go-internal v1.10.0/go.mod h1:UQnix2H7Ngw/k4C5ijL5+65zddjncjaFoBhdsK/akog=
+github.com/rogpeppe/go-internal v1.11.0 h1:cWPaGQEPrBb5/AsnsZesgZZ9yb1OQ+GOISoDNXVBh4M=
+github.com/rogpeppe/go-internal v1.11.0/go.mod h1:ddIwULY96R17DhadqLgMfk9H9tvdUzkipdSkR5nkCZA=
 github.com/rs/xid v1.4.0/go.mod h1:trrq9SKmegXys3aeAKXMUTdJsYXVwGY3RLcfgqegfbg=
 github.com/rs/zerolog v1.29.0 h1:Zes4hju04hjbvkVkOhdl2HpZa+0PmVwigmo8XoORE5w=
 github.com/rs/zerolog v1.29.0/go.mod h1:NILgTygv/Uej1ra5XxGf82ZFSLk58MFGAUS2o6usyD0=
@@ -612,8 +613,8 @@ github.com/stretchr/testify v1.7.0/go.mod h1:6Fq8oRcR53rry900zMqJjRRixrwX3KX962/
 github.com/stretchr/testify v1.7.1/go.mod h1:6Fq8oRcR53rry900zMqJjRRixrwX3KX962/h/Wwjteg=
 github.com/stretchr/testify v1.8.0/go.mod h1:yNjHg4UonilssWZ8iaSj1OCr/vHnekPRkoO+kdMU+MU=
 github.com/stretchr/testify v1.8.1/go.mod h1:w2LPCIKwWwSfY2zedu0+kehJoqGctiVI29o6fzry7u4=
-github.com/stretchr/testify v1.9.0 h1:HtqpIVDClZ4nwg75+f6Lvsy/wHu+3BoSGCbBAcpTsTg=
-github.com/stretchr/testify v1.9.0/go.mod h1:r2ic/lqez/lEtzL7wO/rwa5dbSLXVDPFyf8C91i36aY=
+github.com/stretchr/testify v1.10.0 h1:Xv5erBjTwe/5IxqUQTdXv5kgmIvbHo3QQyRwhJsOfJA=
+github.com/stretchr/testify v1.10.0/go.mod h1:r2ic/lqez/lEtzL7wO/rwa5dbSLXVDPFyf8C91i36aY=
 github.com/theupdateframework/notary v0.7.0 h1:QyagRZ7wlSpjT5N2qQAh/pN+DVqgekv4DzbAiAiEL3c=
 github.com/theupdateframework/notary v0.7.0/go.mod h1:c9DRxcmhHmVLDay4/2fUYdISnHqbFDGRSlXPO0AhYWw=
 github.com/tidwall/gjson v1.14.0 h1:6aeJ0bzojgWLa82gDQHcx3S0Lr/O51I9bJ5nv6JFx5w=