Files
threshold/internal/daemon/worker/controller_connection.go
T
Jeff Mitchell 602bbcc43e Fix intermittent startup issues on slow systems (#3803)
With a suitably slow system -- say, if you're waiting on a Windows
Terminal to output text inside a virtual machine -- the expected order
of things can be thrown off. One specific way this can happen is that
the worker can attempt to dial and have its dial fail because of the
slowness causing the dial context to time out. If, because of the
slowness, we also haven't released the log gate yet, then the event with
the failure information will be queued, along with the context that was
used.

Unfortunately, in some of these error cases, the context that was used
was the dial context instead of the system base context. In many other
places in the function it was the system base context so this is just a
mismatch, probably from code written at different times.

Normally this wouldn't be a problem as we'd fall back to the underlying
logger, but when we release the log gate, things happen differently:
it's a synchronous function and on error it causes us to abandon
system startup entirely.

This commit fixes the issue with using the incorrect context. It's an
open question whether we should change the behavior around errors when
replaying queued events, falling back to the underlying logger instead
of erroring.
2023-10-05 20:34:57 -04:00

447 lines
15 KiB
Go

// Copyright (c) HashiCorp, Inc.
// SPDX-License-Identifier: BUSL-1.1
package worker
import (
"context"
"crypto/ed25519"
"crypto/tls"
"crypto/x509"
"encoding/base64"
"encoding/json"
"encoding/pem"
"fmt"
"math"
"math/big"
mathrand "math/rand"
"net"
"os"
"path/filepath"
"strings"
"sync/atomic"
"time"
"github.com/hashicorp/boundary/globals"
"github.com/hashicorp/boundary/internal/cmd/base"
"github.com/hashicorp/boundary/internal/daemon/cluster"
"github.com/hashicorp/boundary/internal/daemon/cluster/handlers"
"github.com/hashicorp/boundary/internal/daemon/worker/internal/metric"
"github.com/hashicorp/boundary/internal/errors"
"github.com/hashicorp/boundary/internal/event"
pbs "github.com/hashicorp/boundary/internal/gen/controller/servers/services"
"github.com/hashicorp/boundary/internal/server"
"github.com/hashicorp/boundary/internal/util"
"github.com/hashicorp/boundary/version"
"github.com/hashicorp/go-secure-stdlib/base62"
"github.com/hashicorp/go-secure-stdlib/parseutil"
"github.com/hashicorp/nodeenrollment"
"github.com/hashicorp/nodeenrollment/multihop"
"github.com/hashicorp/nodeenrollment/protocol"
"github.com/hashicorp/nodeenrollment/util/toggledlogger"
"google.golang.org/grpc"
"google.golang.org/grpc/backoff"
"google.golang.org/grpc/connectivity"
"google.golang.org/grpc/credentials/insecure"
"google.golang.org/grpc/resolver"
"google.golang.org/protobuf/proto"
"google.golang.org/protobuf/types/known/structpb"
)
var HandleHcpbClusterId func(s string) string
// StartControllerConnections starts up the resolver and initiates controller
// connection client creation
func (w *Worker) StartControllerConnections() error {
const op = "worker.(Worker).StartControllerConnections"
initialAddrs := make([]string, 0, len(w.conf.RawConfig.Worker.InitialUpstreams))
for _, addr := range w.conf.RawConfig.Worker.InitialUpstreams {
switch {
case strings.HasPrefix(addr, "/"):
initialAddrs = append(initialAddrs, addr)
default:
host, port, err := net.SplitHostPort(addr)
if err != nil && strings.Contains(err.Error(), globals.MissingPortErrStr) {
host, port, err = net.SplitHostPort(net.JoinHostPort(addr, "9201"))
}
if err != nil {
return fmt.Errorf("error parsing upstream address: %w", err)
}
initialAddrs = append(initialAddrs, net.JoinHostPort(host, port))
}
}
if len(initialAddrs) == 0 {
if w.conf.RawConfig.HcpbClusterId != "" && HandleHcpbClusterId != nil {
clusterId, err := parseutil.ParsePath(w.conf.RawConfig.HcpbClusterId)
if err != nil && !errors.Is(err, parseutil.ErrNotAUrl) {
return fmt.Errorf("failed to parse HCP Boundary cluster ID: %q: %w", clusterId, err)
}
clusterAddress := HandleHcpbClusterId(clusterId)
initialAddrs = append(initialAddrs, clusterAddress)
event.WriteSysEvent(w.baseContext, op, fmt.Sprintf("Setting HCP Boundary cluster address %s as upstream address", clusterAddress))
} else {
return errors.New(w.baseContext, errors.InvalidParameter, op, "no initial upstream addresses found")
}
}
for _, ar := range w.addressReceivers {
ar.InitialAddresses(initialAddrs)
}
if err := w.createClientConn(initialAddrs[0]); err != nil {
return fmt.Errorf("error making client connection to upstream address %s: %w", initialAddrs[0], err)
}
return nil
}
// upstreamDialerFunc dials an upstream server. extraAlpnProtos can be provided
// to kms and pki connections and are used for identifying, on the server side,
// the intended purpose of the connection. upstreamDialerFunc takes an optional
// stateProvidingFunction which, if the connection is made using the nodeenrollment
// library, sends the state provided by it to the server and can be retrieved
// from the resulting *protocol.Conn.
func (w *Worker) upstreamDialerFunc(extraAlpnProtos ...string) func(context.Context, string) (net.Conn, error) {
const op = "worker.(Worker).upstreamDialerFunc"
return func(ctx context.Context, addr string) (net.Conn, error) {
var conn net.Conn
eventLogger, err := event.NewHclogLogger(w.baseContext, w.conf.Eventer)
if err != nil {
event.WriteError(w.baseContext, op, err)
return nil, errors.Wrap(w.baseContext, err, op)
}
// Give the log a prefix
eventLogger = eventLogger.Named(fmt.Sprintf("workerauth_dialer"))
// Wrap the log in a toggle so we can turn it on and off via config and
// SIGHUP
eventLogger = toggledlogger.NewToggledLogger(eventLogger, w.conf.WorkerAuthDebuggingEnabled)
st, err := w.workerConnectionInfo(addr)
if err != nil {
event.WriteError(w.baseContext, op, err)
return nil, errors.Wrap(w.baseContext, err, op)
}
switch {
case w.conf.RawConfig.Worker.UseDeprecatedKmsAuthMethod:
conn, err = w.v1KmsAuthDialFn(ctx, addr, extraAlpnProtos...)
default:
conn, err = protocol.Dial(
ctx,
w.WorkerAuthStorage,
addr,
nodeenrollment.WithLogger(eventLogger),
nodeenrollment.WithState(st),
nodeenrollment.WithStorageWrapper(w.conf.WorkerAuthStorageKms),
nodeenrollment.WithRegistrationWrapper(w.conf.WorkerAuthKms),
nodeenrollment.WithWrappingRegistrationFlowApplicationSpecificParams(st),
nodeenrollment.WithExtraAlpnProtos(extraAlpnProtos),
// If the activation token hasn't been populated, this won't do
// anything, and it won't do anything if it's already been used
nodeenrollment.WithActivationToken(w.conf.RawConfig.Worker.ControllerGeneratedActivationToken),
)
// No error and a valid connection means the WorkerAuthRegistrationRequest was populated
// We can remove the stored workerAuthRequest file
if err == nil && conn != nil {
if w.conf.RawConfig.Worker.AuthStoragePath != "" {
workerAuthReqFilePath := filepath.Join(w.conf.RawConfig.Worker.AuthStoragePath, base.WorkerAuthReqFile)
// Intentionally ignoring any error removing this file
_ = os.Remove(workerAuthReqFilePath)
}
}
}
switch {
case err == nil:
// Nothing
case errors.Is(err, nodeenrollment.ErrNotAuthorized):
switch w.conf.RawConfig.Worker.ControllerGeneratedActivationToken {
case "":
// We don't event in this case, because the function retries
// often and will spam the logs while waiting on the user to
// transfer the worker-generated request over
return nil, errors.Wrap(w.baseContext, err, op)
default:
// In this case, event, so that the operator can understand that
// it was rejected
event.WriteError(w.baseContext, op, fmt.Errorf("controller rejected activation token as invalid"))
return nil, errors.Wrap(w.baseContext, err, op)
}
default:
event.WriteError(w.baseContext, op, err)
return nil, errors.Wrap(w.baseContext, err, op)
}
if conn != nil {
if w.everAuthenticated.Load() == authenticationStatusNeverAuthenticated {
w.everAuthenticated.Store(authenticationStatusFirstAuthentication)
}
event.WriteSysEvent(w.baseContext, op, "worker has successfully authenticated")
}
return conn, err
}
}
func (w *Worker) v1KmsAuthDialFn(ctx context.Context, addr string, extraAlpnProtos ...string) (net.Conn, error) {
const op = "worker.(Worker).v1KmsAuthDialFn"
tlsConf, authInfo, err := w.workerAuthTLSConfig(extraAlpnProtos...)
if err != nil {
return nil, fmt.Errorf("error creating tls config for worker auth: %w", err)
}
dialer := &net.Dialer{}
var nonTlsConn net.Conn
switch {
case strings.HasPrefix(addr, "/"):
nonTlsConn, err = dialer.DialContext(ctx, "unix", addr)
default:
nonTlsConn, err = dialer.DialContext(ctx, "tcp", addr)
}
if err != nil {
return nil, fmt.Errorf("unable to dial to upstream: %w", err)
}
tlsConn := tls.Client(nonTlsConn, tlsConf)
written, err := tlsConn.Write([]byte(authInfo.ConnectionNonce))
if err != nil {
if err := nonTlsConn.Close(); err != nil {
event.WriteError(w.baseContext, op, err, event.WithInfoMsg("error closing connection after writing failure"))
}
return nil, fmt.Errorf("unable to write connection nonce: %w", err)
}
if written != len(authInfo.ConnectionNonce) {
if err := nonTlsConn.Close(); err != nil {
event.WriteError(w.baseContext, op, err, event.WithInfoMsg("error closing connection after writing failure"))
}
return nil, fmt.Errorf("expected to write %d bytes of connection nonce, wrote %d", len(authInfo.ConnectionNonce), written)
}
return tlsConn, nil
}
func (w *Worker) createClientConn(addr string) error {
const op = "worker.(Worker).createClientConn"
var res resolver.Builder
for _, v := range w.addressReceivers {
if rec, ok := v.(*grpcResolverReceiver); ok {
res = rec.Resolver
}
}
if res == nil {
return errors.New(w.baseContext, errors.Internal, op, "unable to find a resolver.Builder amongst the address receivers")
}
dialOpts := createDefaultGRPCDialOptions(res, w.upstreamDialerFunc())
cc, err := grpc.DialContext(w.baseContext,
fmt.Sprintf("%s:///%s", res.Scheme(), addr),
dialOpts...,
)
if err != nil {
return fmt.Errorf("error dialing controller for worker auth: %w", err)
}
w.GrpcClientConn = cc
w.controllerMultihopConn.Store(multihop.NewMultihopServiceClient(cc))
var producer handlers.UpstreamMessageServiceClientProducer
producer = func(context.Context) (pbs.UpstreamMessageServiceClient, error) {
return pbs.NewUpstreamMessageServiceClient(cc), nil
}
w.controllerUpstreamMsgConn.Store(&producer)
go monitorUpstreamConnectionState(w.baseContext, cc, w.upstreamConnectionState)
return nil
}
// createDefaultGRPCDialOptions creates grpc.DialOption using default options
func createDefaultGRPCDialOptions(res resolver.Builder, upstreamDialerFn func(context.Context, string) (net.Conn, error)) []grpc.DialOption {
defaultTimeout := (time.Second + time.Nanosecond).String()
defServiceConfig := fmt.Sprintf(`
{
"loadBalancingConfig": [ { "round_robin": {} } ],
"methodConfig": [
{
"name": [],
"timeout": %q,
"waitForReady": true
}
]
}
`, defaultTimeout)
dialOpts := []grpc.DialOption{
grpc.WithResolvers(res),
grpc.WithUnaryInterceptor(metric.InstrumentClusterClient()),
grpc.WithDefaultCallOptions(grpc.MaxCallRecvMsgSize(math.MaxInt32)),
grpc.WithDefaultCallOptions(grpc.MaxCallSendMsgSize(math.MaxInt32)),
grpc.WithContextDialer(upstreamDialerFn),
grpc.WithTransportCredentials(insecure.NewCredentials()),
grpc.WithDefaultServiceConfig(defServiceConfig),
// Don't have the resolver reach out for a service config from the
// resolver, use the one specified as default
grpc.WithDisableServiceConfig(),
grpc.WithConnectParams(grpc.ConnectParams{
Backoff: backoff.Config{
BaseDelay: time.Second,
Multiplier: 1.2,
Jitter: 0.2,
MaxDelay: 3 * time.Second,
},
}),
}
return dialOpts
}
func (w *Worker) workerAuthTLSConfig(extraAlpnProtos ...string) (*tls.Config, *base.WorkerAuthInfo, error) {
var err error
info := &base.WorkerAuthInfo{
Name: w.conf.RawConfig.Worker.Name,
Description: w.conf.RawConfig.Worker.Description,
ProxyAddress: w.conf.RawConfig.Worker.PublicAddr,
BoundaryVersion: version.Get().VersionNumber(),
}
info.ConnectionNonce, err = w.nonceFn(20)
if err != nil {
return nil, nil, err
}
pubKey, privKey, err := ed25519.GenerateKey(w.conf.SecureRandomReader)
if err != nil {
return nil, nil, err
}
host, err := base62.Random(20)
if err != nil {
return nil, nil, err
}
template := &x509.Certificate{
ExtKeyUsage: []x509.ExtKeyUsage{
x509.ExtKeyUsageServerAuth,
x509.ExtKeyUsageClientAuth,
},
DNSNames: []string{host},
KeyUsage: x509.KeyUsageDigitalSignature | x509.KeyUsageKeyEncipherment | x509.KeyUsageKeyAgreement | x509.KeyUsageCertSign,
SerialNumber: big.NewInt(mathrand.Int63()),
NotBefore: time.Now().Add(-30 * time.Second),
NotAfter: time.Now().Add(globals.WorkerAuthNonceValidityPeriod),
BasicConstraintsValid: true,
IsCA: true,
}
certBytes, err := x509.CreateCertificate(w.conf.SecureRandomReader, template, template, pubKey, privKey)
if err != nil {
return nil, nil, err
}
certPEMBlock := &pem.Block{
Type: "CERTIFICATE",
Bytes: certBytes,
}
info.CertPEM = pem.EncodeToMemory(certPEMBlock)
marshaledKey, err := x509.MarshalPKCS8PrivateKey(privKey)
if err != nil {
return nil, nil, err
}
keyPEMBlock := &pem.Block{
Type: "PRIVATE KEY",
Bytes: marshaledKey,
}
info.KeyPEM = pem.EncodeToMemory(keyPEMBlock)
// Marshal and encrypt
marshaledInfo, err := json.Marshal(info)
if err != nil {
return nil, nil, err
}
encInfo, err := w.conf.WorkerAuthKms.Encrypt(w.baseContext, marshaledInfo)
if err != nil {
return nil, nil, err
}
marshaledEncInfo, err := proto.Marshal(encInfo)
if err != nil {
return nil, nil, err
}
b64alpn := base64.RawStdEncoding.EncodeToString(marshaledEncInfo)
var nextProtos []string
nextProtos = append(nextProtos, extraAlpnProtos...)
var count int
for i := 0; i < len(b64alpn); i += 230 {
end := i + 230
if end > len(b64alpn) {
end = len(b64alpn)
}
nextProtos = append(nextProtos, fmt.Sprintf("v1workerauth-%02d-%s", count, b64alpn[i:end]))
count++
}
cert, err := x509.ParseCertificate(certBytes)
if err != nil {
return nil, nil, err
}
// Build local tls config
rootCAs := x509.NewCertPool()
rootCAs.AddCert(cert)
tlsCert, err := tls.X509KeyPair(info.CertPEM, info.KeyPEM)
if err != nil {
return nil, nil, err
}
tlsConfig := &tls.Config{
ServerName: host,
Certificates: []tls.Certificate{tlsCert},
RootCAs: rootCAs,
NextProtos: nextProtos,
MinVersion: tls.VersionTLS13,
}
return tlsConfig, info, nil
}
// workerConnectionInfo returns the worker's cluster.WorkerConnectionInfo as
// a struct suitable to be used as the state field in a *protocol.Conn
func (w *Worker) workerConnectionInfo(addr string) (*structpb.Struct, error) {
const op = "worker.(Worker).workerConnectionInfo"
wci := &cluster.WorkerConnectionInfo{
UpstreamAddress: addr,
BoundaryVersion: version.Get().VersionNumber(),
Name: w.conf.RawConfig.Worker.Name,
Description: w.conf.RawConfig.Worker.Description,
PublicAddr: w.conf.RawConfig.Worker.PublicAddr,
OperationalState: w.operationalState.Load().(server.OperationalState).String(),
}
if w.LastStatusSuccess() != nil && len(w.LastStatusSuccess().GetWorkerId()) > 0 {
// even though we wont have the worker the first time we dial, any
// redial attempts should result in the worker id being populated
wci.WorkerId = w.LastStatusSuccess().GetWorkerId()
}
st, err := wci.AsConnectionStateStruct()
if err != nil {
return nil, errors.Wrap(w.baseContext, err, op, errors.WithMsg("getting worker state"))
}
return st, nil
}
// monitorUpstreamConnectionState listens for new state changes from grpc client
// connection and updates the state
func monitorUpstreamConnectionState(ctx context.Context, cc *grpc.ClientConn, connectionState *atomic.Value) {
var state connectivity.State
if v := connectionState.Load(); !util.IsNil(v) {
state = v.(connectivity.State)
}
for cc.WaitForStateChange(ctx, state) {
newState := cc.GetState()
// if the client is shutdown, exit function
if newState == connectivity.Shutdown {
return
}
connectionState.Store(newState)
}
}