Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

[NPM] Use IMDSv2 for vpc_id (network_id) lookups #29027

Merged
merged 18 commits into from
Sep 6, 2024
Merged

[NPM] Use IMDSv2 for vpc_id (network_id) lookups #29027

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
18 commits
Select commit Hold shift + click to select a range
b9c6c0a
temp
akarpz Sep 3, 2024
94728c7
fallback to imdsv2 in rootnetns
akarpz Sep 4, 2024
a616c17
fix interface
akarpz Sep 4, 2024
8fbda8f
add unsupported definition
akarpz Sep 4, 2024
1ed62c8
add mock
akarpz Sep 4, 2024
3eb24a6
fix linter
akarpz Sep 4, 2024
22b05de
linter error
akarpz Sep 5, 2024
e1d7907
run mockery
akarpz Sep 5, 2024
56eaed4
Merge branch 'main' into akarpowich/imds_v2_vpc_endpoint
akarpz Sep 5, 2024
3fbb437
update logs
akarpz Sep 5, 2024
01782ae
fix unix url
akarpz Sep 5, 2024
28ad85d
add definition to windows build
akarpz Sep 5, 2024
434c092
force v2 lookup and improve error logs
akarpz Sep 5, 2024
5076fcd
add stub for windows
akarpz Sep 5, 2024
16688da
fix tests
akarpz Sep 6, 2024
78fb2b0
fix typo
akarpz Sep 6, 2024
41c00ea
fix test
akarpz Sep 6, 2024
3e0bc69
various code review improvements
akarpz Sep 6, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
8 changes: 5 additions & 3 deletions cmd/process-agent/subcommands/check/check.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -182,10 +182,12 @@ func RunCheckCmd(deps Dependencies) error {
names = append(names, ch.Name())

_, processModuleEnabled := deps.Syscfg.SysProbeObject().EnabledModules[sysconfig.ProcessModule]
_, networkTracerModuleEnabled := deps.Syscfg.SysProbeObject().EnabledModules[sysconfig.NetworkTracerModule]
cfg := &checks.SysProbeConfig{
MaxConnsPerMessage: deps.Syscfg.SysProbeObject().MaxConnsPerMessage,
SystemProbeAddress: deps.Syscfg.SysProbeObject().SocketAddress,
ProcessModuleEnabled: processModuleEnabled,
MaxConnsPerMessage: deps.Syscfg.SysProbeObject().MaxConnsPerMessage,
SystemProbeAddress: deps.Syscfg.SysProbeObject().SocketAddress,
ProcessModuleEnabled: processModuleEnabled,
NetworkTracerModuleEnabled: networkTracerModuleEnabled,
}

if !matchingCheck(deps.CliParams.checkName, ch) {
Expand Down
11 changes: 11 additions & 0 deletions cmd/system-probe/modules/network_tracer.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -12,6 +12,7 @@ import (
"encoding/json"
"errors"
"fmt"
"io"
"net/http"
"os"
"runtime"
Expand Down Expand Up @@ -108,6 +109,16 @@ func (nt *networkTracer) Register(httpMux *module.Router) error {
logRequests(id, count, len(cs.Conns), start)
}))

httpMux.HandleFunc("/network_id", utils.WithConcurrencyLimit(utils.DefaultMaxConcurrentRequests, func(w http.ResponseWriter, req *http.Request) {
id, err := nt.tracer.GetNetworkID(req.Context())
if err != nil {
log.Errorf("unable to retrieve network_id: %s", err)
w.WriteHeader(500)
return
}
io.WriteString(w, id)
}))

httpMux.HandleFunc("/register", utils.WithConcurrencyLimit(utils.DefaultMaxConcurrentRequests, func(w http.ResponseWriter, req *http.Request) {
id := getClientID(req)
err := nt.tracer.RegisterClient(id)
Expand Down
15 changes: 15 additions & 0 deletions pkg/network/tracer/tracer.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -40,6 +40,7 @@ import (
"github.com/DataDog/datadog-agent/pkg/process/util"
timeresolver "github.com/DataDog/datadog-agent/pkg/security/resolvers/time"
"github.com/DataDog/datadog-agent/pkg/telemetry"
"github.com/DataDog/datadog-agent/pkg/util/ec2"
"github.com/DataDog/datadog-agent/pkg/util/kernel"
"github.com/DataDog/datadog-agent/pkg/util/log"
)
Expand Down Expand Up @@ -850,3 +851,17 @@ func newUSMMonitor(c *config.Config, tracer connection.Tracer) *usm.Monitor {

return monitor
}

// GetNetworkID retrieves the vpc_id (network_id) from IMDS
func (t *Tracer) GetNetworkID(context context.Context) (string, error) {
id := ""
err := kernel.WithRootNS(kernel.ProcFSRoot(), func() error {
var err error
id, err = ec2.GetNetworkID(context)
return err
})
if err != nil {
return "", err
}
return id, nil
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

We should cache the id here to avoid another call to the metadata endpoint.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I believe either way this will only get called once, since this is hit via the Init() method of the PA check which happens only at startup

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

although if multiple checks are calling this it will indeed be called multiple times, so we can save 2 RPCs by caching this

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

upon further reflection I think I would rather not cache this, because:

  • we are maintaining the same call pattern as before
  • we are only calling max 3 times on startup (if process/container/network checks are all enabled)
  • anecdotally I have seen a failure to fetch the IMDSv2 token sometimes when testing. I am going to follow up with ASC about this, but I am wary to not populate this value if the first request fails

}
5 changes: 5 additions & 0 deletions pkg/network/tracer/tracer_unsupported.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -34,6 +34,11 @@ func (t *Tracer) GetActiveConnections(_ string) (*network.Connections, error) {
return nil, ebpf.ErrNotImplemented
}

// GetNetworkID is not implemented on this OS for Tracer
func (t *Tracer) GetNetworkID(_ context.Context) (string, error) {
return "", ebpf.ErrNotImplemented
}

// RegisterClient registers the client
func (t *Tracer) RegisterClient(_ string) error {
return ebpf.ErrNotImplemented
Expand Down
5 changes: 5 additions & 0 deletions pkg/network/tracer/tracer_windows.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -309,6 +309,11 @@ func (t *Tracer) DebugDumpProcessCache(_ context.Context) (interface{}, error) {
return nil, ebpf.ErrNotImplemented
}

// GetNetworkID is not implemented on this OS for Tracer
func (t *Tracer) GetNetworkID(_ context.Context) (string, error) {
return "", ebpf.ErrNotImplemented
akarpz marked this conversation as resolved.
Show resolved Hide resolved
}

func newUSMMonitor(c *config.Config, dh driver.Handle) usm.Monitor {
if !c.EnableHTTPMonitoring && !c.EnableNativeTLSMonitoring {
return nil
Expand Down
2 changes: 2 additions & 0 deletions pkg/process/checks/checks.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -35,6 +35,8 @@ type SysProbeConfig struct {
SystemProbeAddress string
// System probe process module on/off configuration
ProcessModuleEnabled bool
// System probe network_tracer module on/off configuration
NetworkTracerModuleEnabled bool
}

// Check is an interface for Agent checks that collect data. Each check returns
Expand Down
17 changes: 13 additions & 4 deletions pkg/process/checks/container.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6,7 +6,6 @@
package checks

import (
"context"
"fmt"
"math"
"sync"
Expand All @@ -16,9 +15,9 @@ import (

workloadmeta "github.com/DataDog/datadog-agent/comp/core/workloadmeta/def"
ddconfig "github.com/DataDog/datadog-agent/pkg/config"
"github.com/DataDog/datadog-agent/pkg/process/net"
"github.com/DataDog/datadog-agent/pkg/process/statsd"
proccontainers "github.com/DataDog/datadog-agent/pkg/process/util/containers"
"github.com/DataDog/datadog-agent/pkg/util/cloudproviders"
"github.com/DataDog/datadog-agent/pkg/util/flavor"
"github.com/DataDog/datadog-agent/pkg/util/log"
)
Expand Down Expand Up @@ -53,11 +52,21 @@ type ContainerCheck struct {
}

// Init initializes a ContainerCheck instance.
func (c *ContainerCheck) Init(_ *SysProbeConfig, info *HostInfo, _ bool) error {
func (c *ContainerCheck) Init(syscfg *SysProbeConfig, info *HostInfo, _ bool) error {
c.containerProvider = proccontainers.GetSharedContainerProvider(c.wmeta)
c.hostInfo = info

networkID, err := cloudproviders.GetNetworkID(context.TODO())
var tu *net.RemoteSysProbeUtil
var err error
if syscfg.NetworkTracerModuleEnabled {
// Calling the remote tracer will cause it to initialize and check connectivity
tu, err = net.GetRemoteSystemProbeUtil(syscfg.SystemProbeAddress)
if err != nil {
log.Warnf("could not initiate connection with system probe: %s", err)
}
}

networkID, err := retryGetNetworkID(tu)
if err != nil {
log.Infof("no network ID detected: %s", err)
}
Expand Down
16 changes: 15 additions & 1 deletion pkg/process/checks/net.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -107,7 +107,7 @@ func (c *ConnectionsCheck) Init(syscfg *SysProbeConfig, hostInfo *HostInfo, _ bo
}
}

networkID, err := cloudproviders.GetNetworkID(context.TODO())
networkID, err := retryGetNetworkID(tu)
if err != nil {
log.Infof("no network ID detected: %s", err)
}
Expand Down Expand Up @@ -503,3 +503,17 @@ func convertAndEnrichWithServiceCtx(tags []string, tagOffsets []uint32, serviceC

return tagsStr
}

// fetches network_id from the current netNS or from the system probe if necessary, where the root netNS is used
func retryGetNetworkID(sysProbeUtil *net.RemoteSysProbeUtil) (string, error) {
networkID, err := cloudproviders.GetNetworkID(context.TODO())
if err != nil && sysProbeUtil != nil {
log.Infof("no network ID detected. retrying via system-probe: %s", err)
networkID, err = sysProbeUtil.GetNetworkID()
if err != nil {
log.Infof("failed to get network ID from system-probe: %s", err)
return "", err
}
}
return networkID, err
}
14 changes: 11 additions & 3 deletions pkg/process/checks/process.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6,7 +6,6 @@
package checks

import (
"context"
"errors"
"fmt"
"math"
Expand All @@ -28,7 +27,6 @@ import (
"github.com/DataDog/datadog-agent/pkg/process/statsd"
"github.com/DataDog/datadog-agent/pkg/process/util"
proccontainers "github.com/DataDog/datadog-agent/pkg/process/util/containers"
"github.com/DataDog/datadog-agent/pkg/util/cloudproviders"
"github.com/DataDog/datadog-agent/pkg/util/flavor"
"github.com/DataDog/datadog-agent/pkg/util/log"
"github.com/DataDog/datadog-agent/pkg/util/subscriptions"
Expand Down Expand Up @@ -137,7 +135,17 @@ func (p *ProcessCheck) Init(syscfg *SysProbeConfig, info *HostInfo, oneShot bool

p.notInitializedLogLimit = log.NewLogLimit(1, time.Minute*10)

networkID, err := cloudproviders.GetNetworkID(context.TODO())
var tu *net.RemoteSysProbeUtil
var err error
if syscfg.NetworkTracerModuleEnabled {
// Calling the remote tracer will cause it to initialize and check connectivity
tu, err = net.GetRemoteSystemProbeUtil(syscfg.SystemProbeAddress)
if err != nil {
log.Warnf("could not initiate connection with system probe: %s", err)
}
}

networkID, err := retryGetNetworkID(tu)
if err != nil {
log.Infof("no network ID detected: %s", err)
}
Expand Down
31 changes: 29 additions & 2 deletions pkg/process/net/common.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -44,6 +44,7 @@ type Conn interface {

const (
contentTypeProtobuf = "application/protobuf"
contentTypeJSON = "application/json"
)

var (
Expand Down Expand Up @@ -166,14 +167,40 @@ func (r *RemoteSysProbeUtil) GetConnections(clientID string) (*model.Connections
return conns, nil
}

// GetNetworkID fetches the network_id (vpc_id) from system-probe
func (r *RemoteSysProbeUtil) GetNetworkID() (string, error) {
req, err := http.NewRequest("GET", networkIDURL, nil)
if err != nil {
return "", fmt.Errorf("failed to create request: %w", err)
}

req.Header.Set("Accept", "text/plain")
resp, err := r.httpClient.Do(req)
if err != nil {
return "", fmt.Errorf("failed to execute request: %w", err)
}
defer resp.Body.Close()

if resp.StatusCode != http.StatusOK {
return "", fmt.Errorf("network_id request failed: url: %s, status code: %d", networkIDURL, resp.StatusCode)
}

body, err := io.ReadAll(resp.Body)
if err != nil {
return "", fmt.Errorf("failed to read response body: %w", err)
}

return string(body), nil
akarpz marked this conversation as resolved.
Show resolved Hide resolved
}

// GetPing returns the results of a ping to a host
func (r *RemoteSysProbeUtil) GetPing(clientID string, host string, count int, interval time.Duration, timeout time.Duration) ([]byte, error) {
req, err := http.NewRequest("GET", fmt.Sprintf("%s/%s?client_id=%s&count=%d&interval=%d&timeout=%d", pingURL, host, clientID, count, interval, timeout), nil)
if err != nil {
return nil, err
}

req.Header.Set("Accept", "application/json")
req.Header.Set("Accept", contentTypeJSON)
resp, err := r.httpClient.Do(req)
if err != nil {
return nil, err
Expand Down Expand Up @@ -208,7 +235,7 @@ func (r *RemoteSysProbeUtil) GetTraceroute(clientID string, host string, port ui
return nil, err
}

req.Header.Set("Accept", "application/json")
req.Header.Set("Accept", contentTypeJSON)
resp, err := r.tracerouteClient.Do(req)
if err != nil {
return nil, err
Expand Down
1 change: 1 addition & 0 deletions pkg/process/net/common_linux.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -18,6 +18,7 @@ const (
pingURL = "http://unix/" + string(sysconfig.PingModule) + "/ping/"
tracerouteURL = "http://unix/" + string(sysconfig.TracerouteModule) + "/traceroute/"
connectionsURL = "http://unix/" + string(sysconfig.NetworkTracerModule) + "/connections"
networkIDURL = "http://unix/" + string(sysconfig.NetworkTracerModule) + "/network_id"
procStatsURL = "http://unix/" + string(sysconfig.ProcessModule) + "/stats"
registerURL = "http://unix/" + string(sysconfig.NetworkTracerModule) + "/register"
statsURL = "http://unix/debug/stats"
Expand Down
5 changes: 5 additions & 0 deletions pkg/process/net/common_unsupported.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -40,6 +40,11 @@ func (r *RemoteSysProbeUtil) GetConnections(_ string) (*model.Connections, error
return nil, ErrNotImplemented
}

// GetNetworkID is not supported
func (r *RemoteSysProbeUtil) GetNetworkID() (string, error) {
return "", ErrNotImplemented
}

// GetStats is not supported
func (r *RemoteSysProbeUtil) GetStats() (map[string]interface{}, error) {
return nil, ErrNotImplemented
Expand Down
1 change: 1 addition & 0 deletions pkg/process/net/common_windows.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -15,6 +15,7 @@ import (

const (
connectionsURL = "http://localhost:3333/" + string(sysconfig.NetworkTracerModule) + "/connections"
networkIDURL = "http://unix/" + string(sysconfig.NetworkTracerModule) + "/network_id"
registerURL = "http://localhost:3333/" + string(sysconfig.NetworkTracerModule) + "/register"
languageDetectionURL = "http://localhost:3333/" + string(sysconfig.LanguageDetectionModule) + "/detect"
statsURL = "http://localhost:3333/debug/stats"
Expand Down
28 changes: 28 additions & 0 deletions pkg/process/net/mocks/sys_probe_util.go
View file
Open in desktop

Some generated files are not rendered by default. Learn more about how customized files appear on GitHub.

1 change: 1 addition & 0 deletions pkg/process/net/shared.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -13,4 +13,5 @@ type SysProbeUtil interface {
GetStats() (map[string]interface{}, error)
GetProcStats(pids []int32) (*model.ProcStatsWithPermByPID, error)
Register(clientID string) error
GetNetworkID() (string, error)
}
2 changes: 1 addition & 1 deletion pkg/util/cloudproviders/network.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -30,7 +30,7 @@ func GetNetworkID(ctx context.Context) (string, error) {
return cache.Get[string](
networkIDCacheKey,
func() (string, error) {
// the the id from configuration
// the id from configuration
if networkID := config.Datadog().GetString("network.id"); networkID != "" {
log.Debugf("GetNetworkID: using configured network ID: %s", networkID)
return networkID, nil
Expand Down
Loading
Loading