Compare commits

...

3 Commits

Author SHA1 Message Date
JackDoan 2df43dc218 Add experimental host_query API for local identity lookups
Programs running alongside nebula have no simple way to ask "who is this
vpn address?" when making authorization decisions, e.g. a nebula-aware
webapp that wants to identify an inbound connection by its source
address instead of presenting a login form. The existing surfaces are
the sshd admin interface (not scriptable from app code) and the
lighthouse-only DNS TXT lookup, which returns raw cert JSON over an
awkward transport.

This adds an opt-in `host_query` config section that serves a small
HTTP+JSON API on a unix socket or tcp address, requiring no client
library to consume:

  GET /v1/host?addr=<vpn addr>  identity of the host owning the address
                                (an established peer, or this node).
                                addr may include a port so a server can
                                pass a connection's RemoteAddr through
                                unparsed.
  GET /v1/self                  this node's own identity.

Responses carry the certificate-derived identity only: name, vpn
addresses, networks, unsafe networks, groups, fingerprint, issuer,
validity window, and cert version.

The self-vs-peer lookup logic is shared with the DNS TXT handler via a
new findCertificateForVpnAddr helper, which also swaps the panicking
GetDefaultCertificate call for the nil-returning accessor so a missing
certificate yields an empty answer instead of a crash.

The listener follows the statsServer lifecycle: the whole section is
reloadable via SIGHUP, including moving between socket paths and tcp
addresses. Unix sockets default to mode 0600, stale sockets left by an
unclean exit are removed at bind time, and a non-socket file at the
configured path is never replaced.
2026-07-24 10:06:25 -05:00
Nate Brown 72bf111209 Add an e2e Drop exit type and a roaming recovery measurement (#1819)
smoke-extra / freebsd-amd64 (push) Failing after 15s
smoke-extra / linux-amd64-ipv6disable (push) Failing after 15s
smoke-extra / netbsd-amd64 (push) Failing after 14s
smoke-extra / openbsd-amd64 (push) Failing after 15s
smoke-extra / linux-386 (push) Failing after 16s
smoke / Run multi node smoke test (push) Failing after 1m37s
Build and test / Static checks (push) Successful in 18s
Build and test / Test linux (push) Failing after 58s
Build and test / Test linux-boringcrypto (push) Failing after 2m45s
Build and test / Test linux-pkcs11 (push) Failing after 2m10s
Build and test / Cross-build linux-arm (push) Successful in 3m11s
Build and test / Cross-build linux-mips (push) Successful in 3m53s
Build and test / Cross-build linux-other (push) Successful in 3m16s
Build and test / Cross-build windows (push) Successful in 1m2s
Build and test / Cross-build freebsd (push) Successful in 1m36s
Build and test / Cross-build netbsd (push) Successful in 1m36s
Build and test / Cross-build openbsd (push) Successful in 1m37s
Build and test / Cross-build mobile (push) Successful in 3m23s
smoke-extra / Run windows smoke test (push) Has been cancelled
Build and test / Test macos (push) Has been cancelled
Build and test / Test windows (push) Has been cancelled
Build and test / CI status (push) Has been cancelled
2026-07-23 17:02:02 -05:00
Nate Brown 1617897043 v1.11.0 changelog (#1792)
smoke-extra / freebsd-amd64 (push) Failing after 25s
smoke-extra / linux-amd64-ipv6disable (push) Failing after 13s
smoke-extra / netbsd-amd64 (push) Failing after 13s
smoke-extra / openbsd-amd64 (push) Failing after 11s
smoke-extra / linux-386 (push) Failing after 12s
smoke / Run multi node smoke test (push) Failing after 1m39s
Build and test / Static checks (push) Successful in 2m9s
Build and test / Test linux (push) Failing after 1m3s
Build and test / Test linux-boringcrypto (push) Failing after 2m48s
Build and test / Test linux-pkcs11 (push) Failing after 2m0s
Build and test / Cross-build linux-arm (push) Successful in 3m17s
Build and test / Cross-build linux-mips (push) Successful in 4m5s
Build and test / Cross-build linux-other (push) Successful in 3m23s
Build and test / Cross-build windows (push) Successful in 1m4s
Build and test / Cross-build freebsd (push) Successful in 1m42s
Build and test / Cross-build netbsd (push) Successful in 1m38s
Build and test / Cross-build openbsd (push) Successful in 1m43s
Build and test / Cross-build mobile (push) Successful in 3m34s
smoke-extra / Run windows smoke test (push) Has been cancelled
Build and test / Test macos (push) Has been cancelled
Build and test / Test windows (push) Has been cancelled
Build and test / CI status (push) Has been cancelled
2026-07-23 13:13:45 -05:00
9 changed files with 1133 additions and 24 deletions
+82
View File
@@ -7,6 +7,88 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
## [Unreleased] ## [Unreleased]
## [1.11.0] - 2026-07-23
See the [v1.11.0](https://github.com/slackhq/nebula/milestone/25?closed=1) milestone for a complete list of changes.
### Breaking
- Logging has switched from logrus to Go's structured `slog`. Log output changes: levels are upper case
(`level=INFO`), trace prints as `level=DEBUG-4`, timestamps are always RFC3339Nano and `logging.timestamp_format`
is ignored, and some messages were reworded. Review any log parsing before upgrading. This is also an API break
for embedders, as constructors now take a `*slog.Logger`. (#1672, #1734, #1621)
- `firewall.inbound_action` and `firewall.outbound_action` (used to set reject vs. drop policy) were each being
applied to the opposite direction, that is now corrected. This only affects how blocked packets are answered, not
which packets the firewall allows or denies. If you set either of these you are getting the behavior of the other
one today and likely want to swap them before upgrading. (#1798)
- On Windows, Nebula now installs WFP PERMIT filters for the nebula adapter and the listener port by default. WFP
sits below Windows Defender Firewall, so any WDF inbound rules you rely on for either will no longer apply. Set
`tun.windows_bypass_wdf` and `listen.windows_bypass_wdf` to false to leave WDF in charge. (#1710)
- On Windows, the nebula device is now set to the `private` network category instead of whatever Windows decided,
which is usually `Public`. This makes the host firewall less restrictive on the overlay. Set
`tun.network_category` to `unset` to keep the old behavior. (#1710)
- Reject packets for non-TCP now use ICMP code 13, communication administratively prohibited, instead of code 3,
port unreachable. Anything keying off the old code needs updating. (#1766, #1768)
- The SSH debug server's profiling commands are now confined to `sshd.sandbox_dir`, which defaults to
`$TMP/nebula-debug`. Relative paths resolve inside it and absolute paths outside it are rejected, so anything
scripting `start-cpu-profile`, `save-heap-profile`, or `save-mutex-profile` with a path elsewhere needs the
directory set. The directory is not created for you. (#1622)
### Added
- Sign the Windows release binaries. (#1718)
- Generate IPv6 reject packets, matching the existing IPv4 behavior. (#1766, #1767, #1768)
- Accept `-` in `nebula-cert` to read from stdin or write to stdout. (#1714)
- Search for both `config.yml` and `config.yaml` in service and command line modes. (#1717)
- Add version labels to the Docker/OCI images. (#1772)
- Rebind the listener and re-query lighthouses on macOS when the underlay network changes, so devices moving
between wifi and wired or between networks recover without waiting for dead tunnel detection. Controlled by
`listen.rebind_on_network_change` (default `true`, not reloadable). (#1816)
### Changed
- Reload the firewall when the unsafe networks in the certificate change. (#1719)
- Reconfigure, start, and stop the stats listener on a config reload instead of requiring a restart. (#1670)
- Update a static host's addresses when they change on reload. (#1713)
- Don't require a port on ICMP firewall rules. (#1609)
- Connection track ICMP traffic. (#1602)
- Return `NODATA` instead of `NXDOMAIN` from the DNS server for a name that exists but has no record of the
requested type, so clients that query `AAAA` first (busybox/Alpine) fall through to `A`. (#1668)
- Record the local host's details in the DNS server. (#1716)
- Install Windows unsafe routes as link routes. (#1709)
- Reduce relay handshake log spam, and only log a handshake send error at error level when the remote list
changes. (#1733, #1765, #1810)
- Start, stop, and reload subsystems (DNS, stats, conntrack, ssh, punchy) cleanly without leaking goroutines. (#1640, #1654, #1661, #1667, #1669, #1708, #1806, #1815)
- `Control` is now safe to stop and wait on from any lifecycle state, and a new `Control.Wait` blocks until nebula
has fully stopped and returns the first fatal reader error. Failed starts release the udp sockets and tun fd
instead of leaking them. (#1794)
- Trigger an immediate lighthouse update when reconnecting to or adding a lighthouse instead of waiting for the next update tick. (#1645)
- Bring the Darwin and OpenBSD tun implementations in line with the other BSDs. (#1703)
- Update to build against go v1.26. (#1818)
- Various dependency updates. (#1586, #1587, #1604, #1617, #1618, #1627, #1628, #1629, #1652, #1664, #1665, #1697, #1721, #1732, #1742, #1743, #1750, #1763, #1771, #1782, #1800, #1807)
### Fixed
- Fix a data race on a host's remote address that could send packets to the wrong address during a roam. (#1773)
- Fix tunnels that could permanently escape connection manager monitoring. (#1752)
- Fix a crash when reloading the SSH server's trusted keys. (#1787)
- Fix hostmap corruption when a host has multiple overlay addresses. Each address now gets its own list instead of
a single shared chain, which also fixes two latent bugs on the add and makePrimary paths. (#1788, #1790)
- Apply `remote_allow_list` IPv4 rules to 4-in-6 mapped addresses. (#1786)
- Don't panic in the DNS server on a short or empty query name. (#1635)
- Advance the replay window on relayed packets so a relay drops replayed frames instead of re-forwarding them. (#1751)
- Fix a race in relay state handling. (#1753)
- Lock replay window updates so concurrent readers can't corrupt it. (#1802)
- Reject malformed handshakes more reliably, including invalid ed25519 key lengths. (#1601, #1756)
- Properly handle `closetunnel` packets. (#1638)
- Fix an IPv6 extension-header length overflow that could make the firewall parse the wrong protocol and ports. (#1789)
- Fix relay re-establishment when a handshake arrives over a relay entry that a one-sided teardown left
`Disestablished`, which silently dropped every send until dead tunnel detection forced a re-handshake. (#1805)
- Don't build new relay state on a tunnel that was just discarded. (#1796)
- Don't delete the wrong pending hostinfo in the handshake manager. (#1811)
- Don't call the packet reader after a UDP error on Darwin. (#1755)
- Open the FreeBSD tun device non blocking. (#1666)
## [1.10.3] - 2026-02-06 ## [1.10.3] - 2026-02-06
### Security ### Security
+4
View File
@@ -52,6 +52,7 @@ type Control struct {
sshStart func() sshStart func()
statsStart func() statsStart func()
dnsStart func() dnsStart func()
infoAPIStart func()
lighthouseStart func() lighthouseStart func()
networkChangeStart func(rebind func()) networkChangeStart func(rebind func())
connectionManagerStart func(context.Context) connectionManagerStart func(context.Context)
@@ -108,6 +109,9 @@ func (c *Control) Start() error {
if c.networkChangeStart != nil { if c.networkChangeStart != nil {
go c.networkChangeStart(c.RebindUDPServer) go c.networkChangeStart(c.RebindUDPServer)
} }
if c.infoAPIStart != nil {
go c.infoAPIStart()
}
if c.connectionManagerStart != nil { if c.connectionManagerStart != nil {
go c.connectionManagerStart(c.ctx) go c.connectionManagerStart(c.ctx)
} }
+3 -22
View File
@@ -258,31 +258,12 @@ func (d *dnsServer) QueryCert(data string) string {
return "" return ""
} }
// The hostmap only ever contains peers we have handshaked with, so it never carries an entry for ourselves. crt := findCertificateForVpnAddr(d.certState(), d.hostMap, ip)
// Answer self lookups straight from the local cert state. if crt == nil {
if cs := d.certState(); cs != nil && cs.myVpnAddrsTable != nil && cs.myVpnAddrsTable.Contains(ip) {
c := cs.GetDefaultCertificate()
if c == nil {
return ""
}
b, err := c.MarshalJSON()
if err != nil {
return ""
}
return string(b)
}
hostinfo := d.hostMap.QueryVpnAddr(ip)
if hostinfo == nil {
return "" return ""
} }
q := hostinfo.GetCert() b, err := crt.MarshalJSON()
if q == nil {
return ""
}
b, err := q.Certificate.MarshalJSON()
if err != nil { if err != nil {
return "" return ""
} }
+136
View File
@@ -0,0 +1,136 @@
//go:build e2e_testing
// +build e2e_testing
package e2e
import (
"testing"
"time"
"github.com/slackhq/nebula"
"github.com/slackhq/nebula/cert"
"github.com/slackhq/nebula/cert_test"
"github.com/slackhq/nebula/e2e/router"
"github.com/slackhq/nebula/udp"
)
// TestRecoveryTiming measures how long a tunnel takes to come back after the peer stops accepting our traffic,
// which is what a laptop waking on a new network looks like from the peer's side: its NAT has no state for where
// we are now, so everything we send disappears.
//
// It is a measurement, not a pass/fail assertion. Recovery is timed to the moment the peer punches back at us,
// since that is when its NAT opens and the tunnel is usable again.
//
// go test -tags e2e_testing -v -run TestRecoveryTiming ./e2e/
func TestRecoveryTiming(t *testing.T) {
for _, tc := range []struct {
name string
rebind bool
}{
{"no trigger", false},
{"rebind counter", true},
} {
t.Run(tc.name, func(t *testing.T) {
d, lost := measureRecovery(t, tc.rebind)
t.Logf("RESULT %-16s recovered in %-9v (%d packets lost)", tc.name, d.Round(time.Millisecond), lost)
})
}
}
// measureRecovery returns how long until the peer punched back, and how many of our packets died meanwhile. When
// rebind is true we call RebindUDPServer once the tunnel goes dark, which is what the darwin network change
// monitor does and what iOS has always done. When false, nothing tells nebula anything is wrong.
func measureRecovery(t *testing.T, rebind bool) (time.Duration, int) {
t.Helper()
ca, _, caKey, _ := cert_test.NewTestCaCert(cert.Version2, cert.Curve_CURVE25519, time.Now(), time.Now().Add(10*time.Minute), nil, nil, []string{})
lhControl, lhVpnIpNet, lhUdpAddr, _ := newSimpleServer(cert.Version2, ca, caKey, "lh", "10.128.0.1/24", m{
"lighthouse": m{"am_lighthouse": true},
})
peerCfg := m{
"lighthouse": m{
"hosts": []any{lhVpnIpNet[0].Addr().String()},
"interval": 600,
"local_allow_list": m{
"10.0.0.0/24": true,
"::/0": false,
},
},
"static_host_map": m{
lhVpnIpNet[0].Addr().String(): []any{lhUdpAddr.String()},
},
}
myControl, myVpnIpNet, myUdpAddr, _ := newSimpleServer(cert.Version2, ca, caKey, "me", "10.128.0.2/24", peerCfg)
theirControl, theirVpnIpNet, theirUdpAddr, _ := newSimpleServer(cert.Version2, ca, caKey, "them", "10.128.0.3/24", peerCfg)
r := router.NewR(t, lhControl, myControl, theirControl)
defer r.RenderFlow()
defer func() {
lhControl.Stop()
myControl.Stop()
theirControl.Stop()
}()
lhControl.Start()
myControl.Start()
theirControl.Start()
r.RouteFor(time.Millisecond * 500)
myControl.InjectLightHouseAddr(theirVpnIpNet[0].Addr(), theirUdpAddr)
theirControl.InjectLightHouseAddr(myVpnIpNet[0].Addr(), myUdpAddr)
myControl.InjectTunPacket(BuildTunUDPPacket(theirVpnIpNet[0].Addr(), 80, myVpnIpNet[0].Addr(), 80, []byte("establish")))
r.RouteFor(time.Second)
if myControl.GetHostInfoByVpnAddr(theirVpnIpNet[0].Addr(), false) == nil {
t.Fatal("failed to establish the tunnel we are measuring")
}
r.RouteFor(time.Millisecond * 500)
// From here the peer's NAT has no state for us, everything we send it disappears
start := time.Now()
blackholed := 0
var recovered time.Duration
if rebind {
myControl.RebindUDPServer()
}
// Keep the tun busy the way someone retrying a stalled connection would
stop := make(chan struct{})
defer close(stop)
go func() {
tick := time.NewTicker(time.Millisecond * 200)
defer tick.Stop()
for {
select {
case <-stop:
return
case <-tick.C:
myControl.InjectTunPacket(BuildTunUDPPacket(
theirVpnIpNet[0].Addr(), 80, myVpnIpNet[0].Addr(), 80, []byte("retry")))
}
}
}()
r.RouteForAllExitFuncOrTimeout(time.Second*30, func(p *udp.Packet, c *nebula.Control) router.ExitType {
if c == theirControl && p.From == myControl.GetUDPAddr() {
blackholed++
return router.Drop
}
// The peer reaching us directly is the moment its NAT opened, whether that is a punch or a handshake
if c == myControl && p.From == theirUdpAddr {
recovered = time.Since(start)
return router.RouteAndExit
}
return router.KeepRouting
})
if recovered == 0 {
t.Fatalf("no recovery within 30s (%d packets blackholed)", blackholed)
}
return recovered, blackholed
}
+19 -2
View File
@@ -153,6 +153,9 @@ const (
ExitNow ExitType = 1 ExitNow ExitType = 1
// RouteAndExit routes this packet and exits immediately afterwards // RouteAndExit routes this packet and exits immediately afterwards
RouteAndExit ExitType = 2 RouteAndExit ExitType = 2
// Drop discards this packet without delivering it and keeps routing. Use it to simulate a blackhole, such as
// a restrictive NAT refusing traffic from an address it has not seen.
Drop ExitType = 3
) )
type ExitFunc func(packet *udp.Packet, receiver *nebula.Control) ExitType type ExitFunc func(packet *udp.Packet, receiver *nebula.Control) ExitType
@@ -163,7 +166,9 @@ type ExitFunc func(packet *udp.Packet, receiver *nebula.Control) ExitType
func NewR(t testing.TB, controls ...*nebula.Control) *R { func NewR(t testing.TB, controls ...*nebula.Control) *R {
ctx, cancel := context.WithCancel(context.Background()) ctx, cancel := context.WithCancel(context.Background())
if err := os.MkdirAll("mermaid", 0755); err != nil { // t.Name() contains a slash for subtests, so the flow log can land in a nested directory
fn := filepath.Join("mermaid", fmt.Sprintf("%s.md", t.Name()))
if err := os.MkdirAll(filepath.Dir(fn), 0755); err != nil {
panic(err) panic(err)
} }
@@ -174,7 +179,7 @@ func NewR(t testing.TB, controls ...*nebula.Control) *R {
outNat: make(map[outNatKey]netip.AddrPort), outNat: make(map[outNatKey]netip.AddrPort),
flow: []flowEntry{}, flow: []flowEntry{},
ignoreFlows: []ignoreFlow{}, ignoreFlows: []ignoreFlow{},
fn: filepath.Join("mermaid", fmt.Sprintf("%s.md", t.Name())), fn: fn,
t: t, t: t,
cancelRender: cancel, cancelRender: cancel,
} }
@@ -687,6 +692,10 @@ func (r *R) RouteExitFunc(sender *nebula.Control, whatDo ExitFunc) {
p.Release() p.Release()
return return
case Drop:
// Record it so the flow log shows the attempt, but never hand it to the receiver
r.unlockedInjectFlow(sender, receiver, p, false)
case KeepRouting: case KeepRouting:
fp := r.unlockedInjectFlow(sender, receiver, p, false) fp := r.unlockedInjectFlow(sender, receiver, p, false)
receiver.InjectUDPPacket(p) receiver.InjectUDPPacket(p)
@@ -779,6 +788,10 @@ func (r *R) RouteForAllExitFuncOrTimeout(timeout time.Duration, whatDo ExitFunc)
p.Release() p.Release()
return true return true
case Drop:
// Record it so the flow log shows the attempt, but never hand it to the receiver
r.unlockedInjectFlow(cm[x], receiver, p, false)
case KeepRouting: case KeepRouting:
fp := r.unlockedInjectFlow(cm[x], receiver, p, false) fp := r.unlockedInjectFlow(cm[x], receiver, p, false)
receiver.InjectUDPPacket(p) receiver.InjectUDPPacket(p)
@@ -884,6 +897,10 @@ func (r *R) RouteForAllExitFunc(whatDo ExitFunc) {
p.Release() p.Release()
return return
case Drop:
// Record it so the flow log shows the attempt, but never hand it to the receiver
r.unlockedInjectFlow(cm[x], receiver, p, false)
case KeepRouting: case KeepRouting:
fp := r.unlockedInjectFlow(cm[x], receiver, p, false) fp := r.unlockedInjectFlow(cm[x], receiver, p, false)
receiver.InjectUDPPacket(p) receiver.InjectUDPPacket(p)
+29
View File
@@ -231,6 +231,35 @@ punchy:
# Overriding this to "" is the same as "/" and will allow overwriting any path on the host. # Overriding this to "" is the same as "/" and will allow overwriting any path on the host.
#sandbox_dir: /var/tmp/nebula-debug #sandbox_dir: /var/tmp/nebula-debug
# EXPERIMENTAL: this feature may change or disappear in the future.
# info_api exposes a small local HTTP+JSON API that lets other programs on
# this machine resolve a vpn address to its certificate identity (name, vpn
# addresses, groups, fingerprint, validity), e.g. for making authorization
# decisions about an inbound connection:
# GET /v1/host?addr=<vpn addr> - identity of the host owning the address: a
# peer with an active tunnel, or this node itself. `addr` may include a
# port (`192.168.100.7:54321`), which is ignored, so a connection's remote
# address can be passed through as is. Returns 404 when the address is
# unknown or has no active tunnel.
# GET /v1/self - this node's own identity.
# Identity answers can be trusted because nebula drops inbound packets whose
# source vpn address is not contained in the sender's certificate, so the
# source address of a connection arriving over the nebula interface is
# guaranteed to map to the certificate reported here.
# There is no authentication in this API; restrict access with unix socket
# file permissions.
# This whole section is reloadable.
#info_api:
# Toggles the feature
#enabled: false
# listen accepts a unix socket path as a unix:// URL with an absolute path:
#listen: unix:///var/run/nebula-info-api.sock
# File mode for the unix socket, as an octal string.
# The socket is created by nebula's user; to grant a group of local services
# access, place the socket in a directory with appropriate permissions
# (e.g. a systemd RuntimeDirectory) and relax this to "0660".
#socket_mode: "0600"
# EXPERIMENTAL: relay support for networks that can't establish direct connections. # EXPERIMENTAL: relay support for networks that can't establish direct connections.
relay: relay:
# Relays are a list of Nebula IP's that peers can use to relay packets to me. # Relays are a list of Nebula IP's that peers can use to relay packets to me.
+409
View File
@@ -0,0 +1,409 @@
package nebula
import (
"context"
"encoding/json"
"errors"
"fmt"
"io/fs"
"log/slog"
"net"
"net/http"
"net/netip"
"os"
"path/filepath"
"strconv"
"strings"
"sync"
"sync/atomic"
"time"
"github.com/slackhq/nebula/cert"
"github.com/slackhq/nebula/config"
)
// infoAPIServer is a small http+json listener on a unix socket that lets other
// programs on this machine resolve a vpn address to its certificate identity (name, groups, networks)
// for making authorization decisions. Lifecycle works like statsServer: the constructor wires the
// reload callback, reload records config, Start runs the runtime, Stop tears it down
type infoAPIServer struct {
l *slog.Logger
ctx context.Context
hostMap *HostMap
pki *PKI
// enabled mirrors `info_api.enabled` so callers of Start don't need to know the gating rules
enabled atomic.Bool
runMu sync.Mutex
runCfg *infoAPIConfig
run *infoAPIRuntime // non-nil while a runtime is live
}
// infoAPIRuntime is the live state owned by a single Start invocation. Stop and Start's exit path
// use pointer equality to tell "my runtime" apart from one that replaced it after a reload
type infoAPIRuntime struct {
server *http.Server
listener net.Listener
}
// infoAPIConfig is a snapshot of the info_api config section, comparable with == so reload can
// detect "no change" cheaply
type infoAPIConfig struct {
enabled bool
listen string // raw config value, for error messages
addr string // unix socket path
// file mode applied to the unix socket after bind
socketMode fs.FileMode
}
// newInfoAPIServerFromConfig builds a infoAPIServer and applies the initial config. The reload
// callback is registered first so a SIGHUP can later enable, fix, or disable the listener even if
// the initial config was bad. Nothing binds until Start, so config tests are side effect free.
// A bad config is logged rather than returned: it must not stop nebula from starting, the feature
// just stays disabled until a reload provides a valid config
func newInfoAPIServerFromConfig(ctx context.Context, l *slog.Logger, pki *PKI, hostMap *HostMap, c *config.C) *infoAPIServer {
h := &infoAPIServer{
l: l,
ctx: ctx,
hostMap: hostMap,
pki: pki,
}
c.RegisterReloadCallback(func(c *config.C) {
if err := h.reload(c, false); err != nil {
h.l.Warn("Failed to reload info API from config", "error", err)
}
})
if err := h.reload(c, true); err != nil {
h.l.Warn("Failed to apply info API config; it will stay disabled until the config is fixed and reloaded", "error", err)
}
return h
}
// reload records the latest config. The initial call only records it, Control.Start launches the
// first runtime via infoAPIStart. Later calls reconcile the running listener with the new config:
// enable, disable, or restart when the listen config changed
func (h *infoAPIServer) reload(c *config.C, initial bool) error {
newCfg, err := loadInfoAPIConfig(c)
if err != nil {
return err
}
h.runMu.Lock()
sameCfg := h.runCfg != nil && *h.runCfg == newCfg
h.runCfg = &newCfg
running := h.run != nil
h.runMu.Unlock()
h.enabled.Store(newCfg.enabled)
if initial || sameCfg {
return nil
}
if running {
h.Stop()
}
if newCfg.enabled {
go h.Start()
}
return nil
}
// Start binds the listener from the latest config and serves until Stop is called or ctx fires.
// Safe to call when disabled or already running (both no-op)
func (h *infoAPIServer) Start() {
if !h.enabled.Load() {
return
}
h.runMu.Lock()
if h.ctx.Err() != nil || h.run != nil || h.runCfg == nil {
h.runMu.Unlock()
return
}
cfg := *h.runCfg
ln, err := h.listen(cfg)
if err != nil {
// drop the cached config so a SIGHUP with the same config retries the bind
h.runCfg = nil
h.runMu.Unlock()
h.l.Error("Failed to start info API listener", "listen", cfg.listen, "error", err)
return
}
mux := http.NewServeMux()
mux.HandleFunc("GET /v1/host", h.handleHost)
mux.HandleFunc("GET /v1/self", h.handleSelf)
srv := &http.Server{Handler: mux, ReadHeaderTimeout: 5 * time.Second}
rt := &infoAPIRuntime{server: srv, listener: ln}
h.run = rt
h.runMu.Unlock()
h.l.Info("Starting info API listener", "addr", ln.Addr())
cleanExit := h.serve(srv, ln)
// A Stop that raced our bind shut the server down before Serve could adopt the listener;
// closing it again is harmless and guarantees a unix socket file gets unlinked
_ = ln.Close()
// Clear our runtime only if nothing has replaced it. Stop races through here too but leaves
// h.run == nil, so the pointer check skips
h.runMu.Lock()
if h.run == rt {
h.run = nil
// an error exit leaves runCfg cached as if it were applied, drop it so a SIGHUP with the
// same config re-triggers Start once the user fixes the underlying problem
if !cleanExit {
h.runCfg = nil
}
}
h.runMu.Unlock()
}
// serve runs srv.Serve and ensures ctx cancellation unblocks it. Returns true if the listener
// exited cleanly (Stop, ctx cancellation), false on an unexpected error
func (h *infoAPIServer) serve(srv *http.Server, ln net.Listener) bool {
// ctx cancellation triggers a server shutdown which in turn unblocks Serve, closing `done` on
// exit keeps the watcher from outliving this call
done := make(chan struct{})
go func() {
select {
case <-h.ctx.Done():
shutdownCtx, cancel := context.WithTimeout(context.Background(), 5*time.Second)
defer cancel()
if err := srv.Shutdown(shutdownCtx); err != nil {
h.l.Warn("Failed to shut down info API listener", "error", err)
}
case <-done:
}
}()
defer close(done)
err := srv.Serve(ln)
if err == nil || errors.Is(err, http.ErrServerClosed) {
return true
}
h.l.Error("Info API listener exited", "error", err)
return false
}
// Stop tears down the active runtime, if any. Idempotent
func (h *infoAPIServer) Stop() {
h.runMu.Lock()
rt := h.run
h.run = nil
h.runMu.Unlock()
if rt == nil {
return
}
shutdownCtx, cancel := context.WithTimeout(context.Background(), 5*time.Second)
defer cancel()
if err := rt.server.Shutdown(shutdownCtx); err != nil {
h.l.Warn("Failed to shut down info API listener", "error", err)
}
}
// listen binds the configured unix socket. It also clears a stale socket file left by an unclean
// exit and applies the configured file mode
func (h *infoAPIServer) listen(cfg infoAPIConfig) (net.Listener, error) {
if fi, err := os.Stat(cfg.addr); err == nil {
if fi.Mode()&os.ModeSocket == 0 {
return nil, fmt.Errorf("info_api.listen path %s exists and is not a socket, refusing to replace it", cfg.addr)
}
// a normal shutdown unlinks the socket, so a file here means a previous process exited
// uncleanly, remove it so the bind below can succeed
if err = os.Remove(cfg.addr); err != nil {
return nil, fmt.Errorf("failed to remove stale socket %s: %w", cfg.addr, err)
}
}
ln, err := net.Listen("unix", cfg.addr)
if err != nil {
return nil, err
}
// The socket is briefly live with umask-derived permissions before this chmod lands, tolerated
// because connections accepted in that window still only reach this read-only API
if err = os.Chmod(cfg.addr, cfg.socketMode); err != nil {
_ = ln.Close()
return nil, fmt.Errorf("failed to set mode on socket %s: %w", cfg.addr, err)
}
return ln, nil
}
func (h *infoAPIServer) certState() *CertState {
if h.pki == nil {
return nil
}
return h.pki.getCertState()
}
// handleHost serves GET /v1/host?addr=<vpn addr>, answering with the identity of the host that
// owns the address: a peer with an active tunnel, or this node itself. addr may include a port,
// which is ignored, so clients can pass a connection's remote address through without parsing it
func (h *infoAPIServer) handleHost(w http.ResponseWriter, r *http.Request) {
q := r.URL.Query().Get("addr")
if q == "" {
writeJSONError(w, http.StatusBadRequest, "missing addr parameter")
return
}
ip, err := parseQueryAddrParam(q)
if err != nil {
writeJSONError(w, http.StatusBadRequest, "invalid address")
return
}
crt := findCertificateForVpnAddr(h.certState(), h.hostMap, ip)
if crt == nil {
writeJSONError(w, http.StatusNotFound, "no active tunnel for address")
return
}
h.writeHostIdentity(w, crt)
}
// handleSelf serves GET /v1/self, answering with this node's own identity
func (h *infoAPIServer) handleSelf(w http.ResponseWriter, r *http.Request) {
var crt cert.Certificate
if cs := h.certState(); cs != nil {
crt = cs.getCertificate(cs.initiatingVersion)
}
if crt == nil {
writeJSONError(w, http.StatusInternalServerError, "no certificate available")
return
}
h.writeHostIdentity(w, crt)
}
func (h *infoAPIServer) writeHostIdentity(w http.ResponseWriter, crt cert.Certificate) {
id, err := newHostIdentity(crt)
if err != nil {
writeJSONError(w, http.StatusInternalServerError, "failed to fingerprint certificate")
return
}
w.Header().Set("Content-Type", "application/json")
if err = json.NewEncoder(w).Encode(id); err != nil {
h.l.Debug("Failed to write info API response", "error", err)
}
}
// findCertificateForVpnAddr answers "who owns this vpn address": ourselves (from local cert state,
// the hostmap never carries an entry for this node) or a peer with an active tunnel. Returns nil
// when the address is unknown or the tunnel is mid-teardown
func findCertificateForVpnAddr(cs *CertState, hostMap *HostMap, ip netip.Addr) cert.Certificate {
if cs != nil && cs.myVpnAddrsTable != nil && cs.myVpnAddrsTable.Contains(ip) {
return cs.getCertificate(cs.initiatingVersion)
}
hostinfo := hostMap.QueryVpnAddr(ip)
if hostinfo == nil {
return nil
}
cc := hostinfo.GetCert()
if cc == nil {
return nil
}
return cc.Certificate
}
// hostIdentity is the json document served for both /v1/host and /v1/self, every field is derived
// from the authenticated certificate alone
type hostIdentity struct {
Name string `json:"name"`
VpnAddrs []netip.Addr `json:"vpnAddrs"`
Networks []netip.Prefix `json:"networks"`
UnsafeNetworks []netip.Prefix `json:"unsafeNetworks"`
Groups []string `json:"groups"`
Fingerprint string `json:"fingerprint"`
Issuer string `json:"issuer"`
NotBefore time.Time `json:"notBefore"`
NotAfter time.Time `json:"notAfter"`
CertVersion int `json:"certVersion"`
}
func newHostIdentity(crt cert.Certificate) (hostIdentity, error) {
fp, err := crt.Fingerprint()
if err != nil {
return hostIdentity{}, err
}
// slices are always allocated so they marshal as [] rather than null
networks := crt.Networks()
id := hostIdentity{
Name: crt.Name(),
VpnAddrs: make([]netip.Addr, 0, len(networks)),
Networks: append(make([]netip.Prefix, 0, len(networks)), networks...),
UnsafeNetworks: append(make([]netip.Prefix, 0, len(crt.UnsafeNetworks())), crt.UnsafeNetworks()...),
Groups: append(make([]string, 0, len(crt.Groups())), crt.Groups()...),
Fingerprint: fp,
Issuer: crt.Issuer(),
NotBefore: crt.NotBefore(),
NotAfter: crt.NotAfter(),
CertVersion: int(crt.Version()),
}
for _, n := range networks {
id.VpnAddrs = append(id.VpnAddrs, n.Addr())
}
return id, nil
}
func writeJSONError(w http.ResponseWriter, status int, msg string) {
w.Header().Set("Content-Type", "application/json")
w.WriteHeader(status)
_ = json.NewEncoder(w).Encode(map[string]string{"error": msg})
}
// parseQueryAddrParam parses the addr query parameter, accepting a bare address or an address with
// a port (`192.168.100.7:54321`, `[fd00::1]:443`) so callers can pass a connection's RemoteAddr
// straight through. The result is unmapped, 4in6 addresses (::ffff:a.b.c.d) become ipv4
func parseQueryAddrParam(s string) (netip.Addr, error) {
if ip, err := netip.ParseAddr(s); err == nil {
return ip.Unmap(), nil
}
ap, err := netip.ParseAddrPort(s)
if err != nil {
return netip.Addr{}, err
}
return ap.Addr().Unmap(), nil
}
func loadInfoAPIConfig(c *config.C) (infoAPIConfig, error) {
cfg := infoAPIConfig{
enabled: c.GetBool("info_api.enabled", false),
listen: c.GetString("info_api.listen", ""),
}
if !cfg.enabled {
return cfg, nil
}
if cfg.listen == "" {
return cfg, errors.New("info_api.listen can not be empty when info_api is enabled")
}
addr, err := parseInfoAPIListen(cfg.listen)
if err != nil {
return cfg, err
}
cfg.addr = addr
// read as a string so yaml can't reinterpret the octal literal
modeStr := c.GetString("info_api.socket_mode", "0600")
mode, err := strconv.ParseUint(modeStr, 8, 32)
if err != nil || fs.FileMode(mode)&^fs.ModePerm != 0 {
return cfg, fmt.Errorf("info_api.socket_mode was not a valid octal file mode: %s", modeStr)
}
cfg.socketMode = fs.FileMode(mode)
return cfg, nil
}
// parseInfoAPIListen extracts the unix socket path from the info_api.listen config value, which
// must be a `unix://` URL with an absolute path, e.g. `unix:///var/run/nebula.sock`
func parseInfoAPIListen(listen string) (addr string, err error) {
path, ok := strings.CutPrefix(listen, "unix://")
if !ok {
return "", fmt.Errorf("info_api.listen must be a unix:// socket path: %s", listen)
} else if !filepath.IsAbs(path) {
return "", fmt.Errorf("info_api.listen unix socket path must be absolute: %s", listen)
}
return path, nil
}
+448
View File
@@ -0,0 +1,448 @@
package nebula
import (
"context"
"encoding/json"
"fmt"
"io/fs"
"log/slog"
"net"
"net/http"
"net/http/httptest"
"net/netip"
"net/url"
"os"
"path/filepath"
"runtime"
"testing"
"time"
"github.com/slackhq/nebula/cert"
"github.com/slackhq/nebula/cert_test"
"github.com/slackhq/nebula/config"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
)
func Test_parseInfoAPIListen(t *testing.T) {
type testCase struct {
listen string
addr string
wantErr bool
}
tests := []testCase{
{listen: "", wantErr: true},
{listen: "unix://", wantErr: true},
{listen: "unix://relative/path.sock", wantErr: true},
{listen: "not an address", wantErr: true},
// tcp host:port addresses are no longer accepted
{listen: "127.0.0.1:8085", wantErr: true},
{listen: "[::1]:8085", wantErr: true},
{listen: "localhost:8085", wantErr: true},
}
// A unix socket path must be absolute for the OS that will bind it, and filepath.IsAbs is
// GOOS-specific. CI runs the suite separately on each OS, so assert the platform's own native
// absolute path is accepted while the other platform's is rejected.
posixPath := "unix:///var/run/nebula.sock"
winPath := `unix://C:\nebula\hq.sock`
if runtime.GOOS == "windows" {
tests = append(tests,
testCase{listen: winPath, addr: `C:\nebula\hq.sock`},
testCase{listen: posixPath, wantErr: true},
)
} else {
tests = append(tests,
testCase{listen: posixPath, addr: "/var/run/nebula.sock"},
testCase{listen: winPath, wantErr: true},
)
}
for _, tt := range tests {
addr, err := parseInfoAPIListen(tt.listen)
if tt.wantErr {
require.Error(t, err, "listen=%q", tt.listen)
continue
}
require.NoError(t, err, "listen=%q", tt.listen)
assert.Equal(t, tt.addr, addr, "listen=%q", tt.listen)
}
}
func Test_loadInfoAPIConfig(t *testing.T) {
c := config.NewC(nil)
// the listen path must be absolute for the OS running the test (CI is per-OS)
listen, wantAddr := "unix:///tmp/hq.sock", "/tmp/hq.sock"
if runtime.GOOS == "windows" {
listen, wantAddr = `unix://C:\tmp\hq.sock`, `C:\tmp\hq.sock`
}
// absent section means disabled, no error
cfg, err := loadInfoAPIConfig(c)
require.NoError(t, err)
assert.False(t, cfg.enabled)
// enabled without a listen address is an error
setInfoAPIConfig(c, true, "", "")
_, err = loadInfoAPIConfig(c)
require.Error(t, err)
// a unix socket gets the default mode
setInfoAPIConfig(c, true, listen, "")
cfg, err = loadInfoAPIConfig(c)
require.NoError(t, err)
assert.Equal(t, wantAddr, cfg.addr)
assert.Equal(t, fs.FileMode(0o600), cfg.socketMode)
setInfoAPIConfig(c, true, listen, "0660")
cfg, err = loadInfoAPIConfig(c)
require.NoError(t, err)
assert.Equal(t, fs.FileMode(0o660), cfg.socketMode)
setInfoAPIConfig(c, true, listen, "withers")
_, err = loadInfoAPIConfig(c)
require.Error(t, err)
// mode bits beyond the permission bits are rejected
setInfoAPIConfig(c, true, listen, "10600")
_, err = loadInfoAPIConfig(c)
require.Error(t, err)
// tcp host:port listen addresses are no longer supported
setInfoAPIConfig(c, true, "127.0.0.1:8085", "")
_, err = loadInfoAPIConfig(c)
require.Error(t, err)
}
func TestInfoAPIServer_badConfigIsNonFatal(t *testing.T) {
// an enabled-but-invalid config must not stop construction; nebula keeps starting and the
// feature simply stays disabled until a reload supplies a valid config
c := config.NewC(nil)
setInfoAPIConfig(c, true, "not-a-unix-socket", "")
h := newInfoAPIServerFromConfig(context.Background(), slog.New(slog.DiscardHandler), nil, newHostMap(slog.New(slog.DiscardHandler)), c)
require.NotNil(t, h)
assert.False(t, h.enabled.Load())
// no config was recorded, so Start has nothing to bind and is a no-op
h.runMu.Lock()
assert.Nil(t, h.runCfg)
h.runMu.Unlock()
h.Start()
h.runMu.Lock()
assert.Nil(t, h.run)
h.runMu.Unlock()
}
func setInfoAPIConfig(c *config.C, enabled bool, listen, socketMode string) {
settings := map[string]any{
"enabled": enabled,
"listen": listen,
}
if socketMode != "" {
settings["socket_mode"] = socketMode
}
c.Settings["info_api"] = settings
}
func newTestInfoAPIServer(t *testing.T) (*infoAPIServer, *config.C) {
t.Helper()
h := &infoAPIServer{
l: slog.New(slog.DiscardHandler),
ctx: context.Background(),
hostMap: newHostMap(slog.New(slog.DiscardHandler)),
}
h.hostMap.preferredRanges.Store(&[]netip.Prefix{})
return h, config.NewC(nil)
}
// addTestPeer creates a certificate for a peer owning each addr (as a /24 or /64) and inserts it
// into the hostmap as an established tunnel
func addTestPeer(t *testing.T, hm *HostMap, name string, addrs []netip.Addr, unsafeNetworks []netip.Prefix, groups []string) cert.Certificate {
t.Helper()
networks := make([]netip.Prefix, 0, len(addrs))
for _, a := range addrs {
bits := 24
if a.Is6() {
bits = 64
}
networks = append(networks, netip.PrefixFrom(a, bits))
}
ca, _, caKey, _ := cert_test.NewTestCaCert(cert.Version2, cert.Curve_CURVE25519, time.Time{}, time.Time{}, nil, nil, nil)
crt, _, _, _ := cert_test.NewTestCert(cert.Version2, cert.Curve_CURVE25519, ca, caKey, name, time.Time{}, time.Time{}, networks, unsafeNetworks, groups)
fp, err := crt.Fingerprint()
require.NoError(t, err)
hm.unlockedAddHostInfo(&HostInfo{
ConnectionState: &ConnectionState{
peerCert: &cert.CachedCertificate{Certificate: crt, Fingerprint: fp},
},
vpnAddrs: addrs,
relayState: RelayState{
relayForByAddr: map[netip.Addr]*Relay{},
relayForByIdx: map[uint32]*Relay{},
},
}, &Interface{})
return crt
}
func getHost(t *testing.T, h *infoAPIServer, addrParam string) (int, map[string]any) {
t.Helper()
r := httptest.NewRequest(http.MethodGet, "/v1/host?addr="+url.QueryEscape(addrParam), nil)
w := httptest.NewRecorder()
h.handleHost(w, r)
return decodeResponse(t, w)
}
func decodeResponse(t *testing.T, w *httptest.ResponseRecorder) (int, map[string]any) {
t.Helper()
assert.Equal(t, "application/json", w.Header().Get("Content-Type"))
var body map[string]any
require.NoError(t, json.Unmarshal(w.Body.Bytes(), &body))
return w.Code, body
}
func TestInfoAPIServer_handleHost(t *testing.T) {
h, _ := newTestInfoAPIServer(t)
h.pki = newTestPKI(t, "self", []netip.Addr{netip.MustParseAddr("10.0.0.1")})
peerV4 := netip.MustParseAddr("10.0.0.99")
peerV6 := netip.MustParseAddr("fd00::99")
addTestPeer(t, h.hostMap, "laptop-alice", []netip.Addr{peerV4, peerV6},
[]netip.Prefix{netip.MustParsePrefix("192.168.50.0/24")}, []string{"eng", "ssh"})
addTestPeer(t, h.hostMap, "groupless", []netip.Addr{netip.MustParseAddr("10.0.0.77")}, nil, nil)
// an established peer comes back with its full identity
code, body := getHost(t, h, "10.0.0.99")
require.Equal(t, http.StatusOK, code)
assert.Equal(t, "laptop-alice", body["name"])
assert.Equal(t, []any{"10.0.0.99", "fd00::99"}, body["vpnAddrs"])
assert.Equal(t, []any{"10.0.0.99/24", "fd00::99/64"}, body["networks"])
assert.Equal(t, []any{"192.168.50.0/24"}, body["unsafeNetworks"])
assert.Equal(t, []any{"eng", "ssh"}, body["groups"])
assert.NotEmpty(t, body["fingerprint"])
assert.Equal(t, "2", fmt.Sprintf("%v", body["certVersion"]))
assert.NotEmpty(t, body["notBefore"])
assert.NotEmpty(t, body["notAfter"])
// empty cert slices marshal as [] rather than null
code, body = getHost(t, h, "10.0.0.77")
require.Equal(t, http.StatusOK, code)
require.NotNil(t, body["groups"])
assert.Empty(t, body["groups"])
require.NotNil(t, body["unsafeNetworks"])
assert.Empty(t, body["unsafeNetworks"])
// a port in addr is ignored so RemoteAddr can be passed through directly, including the
// bracketed v6 and 4in6 forms
for _, q := range []string{"10.0.0.99:54321", "[fd00::99]:443", "::ffff:10.0.0.99"} {
code, body = getHost(t, h, q)
require.Equal(t, http.StatusOK, code, "addr=%q", q)
assert.Equal(t, "laptop-alice", body["name"], "addr=%q", q)
}
// our own address answers from the local cert state
code, body = getHost(t, h, "10.0.0.1")
require.Equal(t, http.StatusOK, code)
assert.Equal(t, "self", body["name"])
code, body = getHost(t, h, "10.0.0.42")
assert.Equal(t, http.StatusNotFound, code)
assert.NotEmpty(t, body["error"])
// a tunnel mid-teardown (no peer cert) is treated as unknown
h.hostMap.unlockedAddHostInfo(&HostInfo{
ConnectionState: &ConnectionState{},
vpnAddrs: []netip.Addr{netip.MustParseAddr("10.0.0.66")},
relayState: RelayState{
relayForByAddr: map[netip.Addr]*Relay{},
relayForByIdx: map[uint32]*Relay{},
},
}, &Interface{})
code, _ = getHost(t, h, "10.0.0.66")
assert.Equal(t, http.StatusNotFound, code)
code, body = getHost(t, h, "not-an-address")
assert.Equal(t, http.StatusBadRequest, code)
assert.NotEmpty(t, body["error"])
r := httptest.NewRequest(http.MethodGet, "/v1/host", nil)
w := httptest.NewRecorder()
h.handleHost(w, r)
code, body = decodeResponse(t, w)
assert.Equal(t, http.StatusBadRequest, code)
assert.NotEmpty(t, body["error"])
}
func TestInfoAPIServer_handleSelf(t *testing.T) {
h, _ := newTestInfoAPIServer(t)
h.pki = newTestPKI(t, "lighthouse", []netip.Addr{netip.MustParseAddr("10.0.0.1")})
r := httptest.NewRequest(http.MethodGet, "/v1/self", nil)
w := httptest.NewRecorder()
h.handleSelf(w, r)
code, body := decodeResponse(t, w)
require.Equal(t, http.StatusOK, code)
assert.Equal(t, "lighthouse", body["name"])
assert.Equal(t, []any{"10.0.0.1"}, body["vpnAddrs"])
// no cert state available should be an error, not a panic
h.pki = nil
w = httptest.NewRecorder()
h.handleSelf(w, r)
code, body = decodeResponse(t, w)
assert.Equal(t, http.StatusInternalServerError, code)
assert.NotEmpty(t, body["error"])
}
func unixHTTPClient(path string) *http.Client {
return &http.Client{
Timeout: time.Second,
Transport: &http.Transport{
DialContext: func(ctx context.Context, _, _ string) (net.Conn, error) {
return (&net.Dialer{}).DialContext(ctx, "unix", path)
},
},
}
}
// waitForServe polls until a GET /v1/self through client succeeds
func waitForServe(t *testing.T, client *http.Client) {
t.Helper()
waitFor(t, func() bool {
resp, err := client.Get("http://hostquery/v1/self")
if err != nil {
return false
}
resp.Body.Close()
return resp.StatusCode == http.StatusOK
})
}
func skipIfNoUnixSockets(t *testing.T) {
t.Helper()
if runtime.GOOS == "windows" {
t.Skip("unix socket tests are not supported on windows CI")
}
}
func TestInfoAPIServer_unixLifecycle(t *testing.T) {
skipIfNoUnixSockets(t)
h, c := newTestInfoAPIServer(t)
h.pki = newTestPKI(t, "self", []netip.Addr{netip.MustParseAddr("10.0.0.1")})
sock := filepath.Join(t.TempDir(), "hq.sock")
setInfoAPIConfig(c, true, "unix://"+sock, "")
require.NoError(t, h.reload(c, true))
done := make(chan struct{})
go func() {
h.Start()
close(done)
}()
client := unixHTTPClient(sock)
waitForServe(t, client)
fi, err := os.Stat(sock)
require.NoError(t, err)
assert.Equal(t, fs.FileMode(0o600), fi.Mode().Perm())
resp, err := client.Get("http://hostquery/v1/host?addr=10.0.0.1")
require.NoError(t, err)
resp.Body.Close()
assert.Equal(t, http.StatusOK, resp.StatusCode)
h.Stop()
select {
case <-done:
case <-time.After(5 * time.Second):
t.Fatal("Start did not return after Stop")
}
_, err = os.Stat(sock)
assert.True(t, os.IsNotExist(err), "socket file should be unlinked on shutdown")
}
func TestInfoAPIServer_staleSocket(t *testing.T) {
skipIfNoUnixSockets(t)
h, _ := newTestInfoAPIServer(t)
sock := filepath.Join(t.TempDir(), "hq.sock")
// simulate an unclean exit, a leftover socket file with no listener
stale, err := net.ListenUnix("unix", &net.UnixAddr{Name: sock, Net: "unix"})
require.NoError(t, err)
stale.SetUnlinkOnClose(false)
require.NoError(t, stale.Close())
_, err = os.Stat(sock)
require.NoError(t, err, "stale socket file should exist")
cfg := infoAPIConfig{addr: sock, socketMode: 0o600}
ln, err := h.listen(cfg)
require.NoError(t, err, "a stale socket should be removed and rebound")
require.NoError(t, ln.Close())
}
func TestInfoAPIServer_existingFileNotReplaced(t *testing.T) {
skipIfNoUnixSockets(t)
h, _ := newTestInfoAPIServer(t)
path := filepath.Join(t.TempDir(), "hq.sock")
require.NoError(t, os.WriteFile(path, []byte("precious"), 0o600))
cfg := infoAPIConfig{addr: path, socketMode: 0o600}
_, err := h.listen(cfg)
require.Error(t, err, "a non-socket file at the listen path must not be replaced")
content, err := os.ReadFile(path)
require.NoError(t, err)
assert.Equal(t, "precious", string(content))
}
func TestInfoAPIServer_reload(t *testing.T) {
skipIfNoUnixSockets(t)
h, c := newTestInfoAPIServer(t)
h.pki = newTestPKI(t, "self", []netip.Addr{netip.MustParseAddr("10.0.0.1")})
dir := t.TempDir()
sock1 := filepath.Join(dir, "hq1.sock")
sock2 := filepath.Join(dir, "hq2.sock")
// initial reload only records config, Control.Start is what launches the runtime
setInfoAPIConfig(c, false, "unix://"+sock1, "")
require.NoError(t, h.reload(c, true))
assert.False(t, h.enabled.Load())
h.runMu.Lock()
assert.Nil(t, h.run)
h.runMu.Unlock()
// enabling via reload spawns the listener
setInfoAPIConfig(c, true, "unix://"+sock1, "")
require.NoError(t, h.reload(c, false))
waitForServe(t, unixHTTPClient(sock1))
// changing the listen path restarts on the new address
setInfoAPIConfig(c, true, "unix://"+sock2, "")
require.NoError(t, h.reload(c, false))
waitForServe(t, unixHTTPClient(sock2))
waitFor(t, func() bool {
_, err := os.Stat(sock1)
return os.IsNotExist(err)
})
// reloading an unchanged config does not restart the runtime
h.runMu.Lock()
rt := h.run
h.runMu.Unlock()
require.NoError(t, h.reload(c, false))
h.runMu.Lock()
assert.Same(t, rt, h.run)
h.runMu.Unlock()
// disabling stops the listener
setInfoAPIConfig(c, false, "unix://"+sock2, "")
require.NoError(t, h.reload(c, false))
assert.False(t, h.enabled.Load())
waitFor(t, func() bool {
h.runMu.Lock()
defer h.runMu.Unlock()
return h.run == nil
})
}
+3
View File
@@ -260,6 +260,8 @@ func Main(c *config.C, configTest bool, buildVersion string, l *slog.Logger, dev
return nil, util.ContextualizeIfNeeded("Failed to start stats emitter", err) return nil, util.ContextualizeIfNeeded("Failed to start stats emitter", err)
} }
infoAPI := newInfoAPIServerFromConfig(ctx, l, pki, hostMap, c)
if configTest { if configTest {
return nil, nil return nil, nil
} }
@@ -279,6 +281,7 @@ func Main(c *config.C, configTest bool, buildVersion string, l *slog.Logger, dev
sshStart: sshStart, sshStart: sshStart,
statsStart: stats.Start, statsStart: stats.Start,
dnsStart: ds.Start, dnsStart: ds.Start,
infoAPIStart: infoAPI.Start,
lighthouseStart: lightHouse.StartUpdateWorker, lighthouseStart: lightHouse.StartUpdateWorker,
networkChangeStart: networkChanges.Start, networkChangeStart: networkChanges.Start,
connectionManagerStart: connManager.Start, connectionManagerStart: connManager.Start,