Drop all the dependencies we had on procspy by moving some code to probe.

This commit is contained in:
Alvaro Saurin
2015-09-01 14:01:30 +02:00
parent b2b8b45bc2
commit 4331fb1c79
19 changed files with 476 additions and 55 deletions

View File

@@ -3,6 +3,7 @@ package docker
import (
"strconv"
"github.com/weaveworks/scope/probe/proc"
"github.com/weaveworks/scope/probe/process"
"github.com/weaveworks/scope/report"
)
@@ -16,18 +17,18 @@ const (
// These vars are exported for testing.
var (
NewProcessTreeStub = process.NewTree
NewProcessTreeStub = proc.NewTree
)
// Tagger is a tagger that tags Docker container information to process
// nodes that have a PID.
type Tagger struct {
registry Registry
procWalker process.Walker
procWalker proc.Walker
}
// NewTagger returns a usable Tagger.
func NewTagger(registry Registry, procWalker process.Walker) *Tagger {
func NewTagger(registry Registry, procWalker proc.Walker) *Tagger {
return &Tagger{
registry: registry,
procWalker: procWalker,
@@ -44,7 +45,7 @@ func (t *Tagger) Tag(r report.Report) (report.Report, error) {
return r, nil
}
func (t *Tagger) tag(tree process.Tree, topology *report.Topology) {
func (t *Tagger) tag(tree proc.Tree, topology *report.Topology) {
for nodeID, nodeMetadata := range topology.Nodes {
pidStr, ok := nodeMetadata.Metadata[process.PID]
if !ok {

View File

@@ -6,7 +6,7 @@ import (
"testing"
"github.com/weaveworks/scope/probe/docker"
"github.com/weaveworks/scope/probe/process"
"github.com/weaveworks/scope/probe/proc"
"github.com/weaveworks/scope/report"
"github.com/weaveworks/scope/test"
)
@@ -27,7 +27,7 @@ func TestTagger(t *testing.T) {
oldProcessTree := docker.NewProcessTreeStub
defer func() { docker.NewProcessTreeStub = oldProcessTree }()
docker.NewProcessTreeStub = func(_ process.Walker) (process.Tree, error) {
docker.NewProcessTreeStub = func(_ proc.Walker) (proc.Tree, error) {
return &mockProcessTree{map[int]int{2: 1}}, nil
}

View File

@@ -7,7 +7,7 @@ import (
"github.com/prometheus/client_golang/prometheus"
"github.com/weaveworks/procspy"
"github.com/weaveworks/scope/probe/proc"
"github.com/weaveworks/scope/probe/process"
"github.com/weaveworks/scope/report"
)
@@ -42,7 +42,7 @@ var SpyDuration = prometheus.NewSummaryVec(
[]string{},
)
// NewReporter creates a new Reporter that invokes procspy.Connections to
// NewReporter creates a new Reporter that invokes Connections to
// generate a report.Report that contains every discovered (spied) connection
// on the host machine, at the granularity of host and port. That information
// is stored in the Endpoint topology. It optionally enriches that topology
@@ -94,7 +94,7 @@ func (r *Reporter) Report() (report.Report, error) {
}(time.Now())
rpt := report.MakeReport()
conns, err := procspy.Connections(r.includeProcesses)
conns, err := proc.Connections(r.includeProcesses)
if err != nil {
return rpt, err
}

View File

@@ -5,7 +5,7 @@ import (
"strconv"
"testing"
"github.com/weaveworks/procspy"
"github.com/weaveworks/scope/probe/proc"
"github.com/weaveworks/scope/probe/docker"
"github.com/weaveworks/scope/probe/endpoint"
"github.com/weaveworks/scope/report"
@@ -20,7 +20,7 @@ var (
fixProcessPID = uint(4242)
fixProcessName = "nginx"
fixConnections = []procspy.Connection{
fixConnections = []proc.Connection{
{
Transport: "tcp",
LocalAddress: fixLocalAddress,
@@ -37,14 +37,14 @@ var (
},
}
fixConnectionsWithProcesses = []procspy.Connection{
fixConnectionsWithProcesses = []proc.Connection{
{
Transport: "tcp",
LocalAddress: fixLocalAddress,
LocalPort: fixLocalPort,
RemoteAddress: fixRemoteAddress,
RemotePort: fixRemotePort,
Proc: procspy.Proc{
Proc: proc.Proc{
PID: fixProcessPID,
Name: fixProcessName,
},
@@ -55,7 +55,7 @@ var (
LocalPort: fixLocalPort,
RemoteAddress: fixRemoteAddress,
RemotePort: fixRemotePort,
Proc: procspy.Proc{
Proc: proc.Proc{
PID: fixProcessPID,
Name: fixProcessName,
},
@@ -64,7 +64,7 @@ var (
)
func TestSpyNoProcesses(t *testing.T) {
procspy.SetFixtures(fixConnections)
proc.SetFixtures(fixConnections)
const (
nodeID = "heinz-tomato-ketchup" // TODO rename to hostID
@@ -100,7 +100,7 @@ func TestSpyNoProcesses(t *testing.T) {
}
func TestSpyWithProcesses(t *testing.T) {
procspy.SetFixtures(fixConnectionsWithProcesses)
proc.SetFixtures(fixConnectionsWithProcesses)
const (
nodeID = "nikon" // TODO rename to hostID

View File

@@ -14,7 +14,7 @@ import (
"syscall"
"time"
"github.com/weaveworks/procspy"
"github.com/weaveworks/scope/probe/proc"
"github.com/weaveworks/scope/probe/docker"
"github.com/weaveworks/scope/probe/endpoint"
"github.com/weaveworks/scope/probe/host"
@@ -67,7 +67,7 @@ func main() {
}
log.Printf("publishing to: %s", strings.Join(targets, ", "))
procspy.SetProcRoot(*procRoot)
proc.SetProcRoot(*procRoot)
if *httpListen != "" {
log.Printf("profiling data being exported to %s", *httpListen)
@@ -112,7 +112,7 @@ func main() {
var (
endpointReporter = endpoint.NewReporter(hostID, hostName, *spyProcs, *useConntrack)
processCache = process.NewCachingWalker(process.NewWalker(*procRoot))
processCache = proc.NewCachingWalker(proc.NewWalker(*procRoot))
tickers = []Ticker{processCache}
reporters = []Reporter{
endpointReporter,

26
probe/proc/fixture.go Normal file
View File

@@ -0,0 +1,26 @@
package proc
// SetFixtures declares constant Connection and ConnectionProcs which will
// always be returned by the package-level Connections and Processes
// functions. It's designed to be used in tests.
type fixedConnIter []Connection
func (f *fixedConnIter) Next() *Connection {
if len(*f) == 0 {
return nil
}
car := (*f)[0]
*f = (*f)[1:]
return &car
}
// SetFixtures is used in test scenarios to have known output.
func SetFixtures(c []Connection) {
cbConnections = func(bool) (ConnIter, error) {
f := fixedConnIter(c)
return &f, nil
}
}

125
probe/proc/proc.go Normal file
View File

@@ -0,0 +1,125 @@
package proc
// /proc-based implementation.
import (
"bytes"
"os"
"strconv"
"syscall"
"path"
)
var (
procRoot = "/proc"
)
// SetProcRoot sets the location of the proc filesystem.
func SetProcRoot(root string) {
procRoot = root
}
// walkProcPid walks over all numerical (PID) /proc entries, and sees if their
// ./fd/* files are symlink to sockets. Returns a map from socket ID (inode)
// to PID. Will return an error if /proc isn't there.
func walkProcPid() (map[uint64]Proc, error) {
fh, err := os.Open(procRoot)
if err != nil {
return nil, err
}
dirNames, err := fh.Readdirnames(-1)
fh.Close()
if err != nil {
return nil, err
}
var (
res = map[uint64]Proc{}
stat syscall.Stat_t
)
for _, dirName := range dirNames {
pid, err := strconv.ParseUint(dirName, 10, 0)
if err != nil {
// Not a number, so not a PID subdir.
continue
}
fdBase := procRoot + "/" + dirName + "/fd/"
dfh, err := os.Open(fdBase)
if err != nil {
// Process is be gone by now, or we don't have access.
continue
}
fdNames, err := dfh.Readdirnames(-1)
dfh.Close()
if err != nil {
continue
}
var name string
for _, fdName := range fdNames {
// Direct use of syscall.Stat() to save garbage.
err = syscall.Stat(fdBase + fdName, &stat)
if err != nil {
continue
}
// We want sockets only.
if stat.Mode & syscall.S_IFMT != syscall.S_IFSOCK {
continue
}
if name == "" {
if name = procName(path.Join(procRoot, dirName)); name == "" {
// Process might be gone by now
break
}
}
res[stat.Ino] = Proc{
PID: uint(pid),
Name: name,
}
}
}
return res, nil
}
// procName does a pid->name lookup.
func procName(base string) string {
fh, err := os.Open(base + "/comm")
if err != nil {
return ""
}
name := make([]byte, 64)
l, err := fh.Read(name)
fh.Close()
if err != nil {
return ""
}
if l < 2 {
return ""
}
// drop trailing "\n"
return string(name[:l - 1])
}
// readFile reads an arbitrary file into a buffer. It's a variable so it can
// be overwritten for benchmarks. That's bad practice and we should change it
// to be a dependency.
var readFile = func(filename string, buf *bytes.Buffer) error {
f, err := os.Open(filename)
if err != nil {
return err
}
_, err = buf.ReadFrom(f)
f.Close()
return err
}

164
probe/proc/procnet.go Normal file
View File

@@ -0,0 +1,164 @@
package proc
import (
"bytes"
"net"
)
// ProcNet is an iterator to parse /proc/net/tcp{,6} files.
type ProcNet struct {
b []byte
c Connection
wantedState uint
bytesLocal, bytesRemote [16]byte
}
// NewProcNet gives a new ProcNet parser.
func NewProcNet(b []byte, wantedState uint) *ProcNet {
return &ProcNet{
b: b,
c: Connection{},
wantedState: wantedState,
}
}
// Next returns the next connection. All buffers are re-used, so if you want
// to keep the IPs you have to copy them.
func (p *ProcNet) Next() *Connection {
again:
if len(p.b) == 0 {
return nil
}
b := p.b
if p.b[2] == 's' {
// Skip header
p.b = nextLine(b)
goto again
}
var (
local, remote, state, inode []byte
)
_, b = nextField(b) // 'sl' column
local, b = nextField(b)
remote, b = nextField(b)
state, b = nextField(b)
if parseHex(state) != p.wantedState {
p.b = nextLine(b)
goto again
}
_, b = nextField(b) // 'tx_queue' column
_, b = nextField(b) // 'rx_queue' column
_, b = nextField(b) // 'tr' column
_, b = nextField(b) // 'uid' column
_, b = nextField(b) // 'timeout' column
inode, b = nextField(b)
p.c.LocalAddress, p.c.LocalPort = scanAddressNA(local, &p.bytesLocal)
p.c.RemoteAddress, p.c.RemotePort = scanAddressNA(remote, &p.bytesRemote)
p.c.inode = parseDec(inode)
p.b = nextLine(b)
return &p.c
}
// scanAddressNA parses 'A12CF62E:00AA' to the address/port. Handles IPv4 and
// IPv6 addresses. The address is a big endian 32 bit ints, hex encoded. We
// just decode the hex and flip the bytes in every group of 4.
func scanAddressNA(in []byte, buf *[16]byte) (net.IP, uint16) {
col := bytes.IndexByte(in, ':')
if col == -1 {
return nil, 0
}
// Network address is big endian. Can be either ipv4 or ipv6.
address := hexDecode32bigNA(in[:col], buf)
return net.IP(address), uint16(parseHex(in[col + 1:]))
}
// hexDecode32big decodes sequences of 32bit big endian bytes.
func hexDecode32bigNA(src []byte, buf *[16]byte) []byte {
blocks := len(src) / 8
for block := 0; block < blocks; block++ {
for i := 0; i < 4; i++ {
a := fromHexChar(src[block * 8 + i * 2])
b := fromHexChar(src[block * 8 + i * 2 + 1])
buf[block * 4 + 3 - i] = (a << 4) | b
}
}
return buf[:blocks * 4]
}
func nextField(s []byte) ([]byte, []byte) {
// Skip whitespace.
for i, b := range s {
if b != ' ' {
s = s[i:]
break
}
}
// Up until the next whitespace field.
for i, b := range s {
if b == ' ' {
return s[:i], s[i:]
}
}
return nil, nil
}
func nextLine(s []byte) []byte {
i := bytes.IndexByte(s, '\n')
if i == -1 {
return nil
}
return s[i + 1:]
}
// Simplified copy of strconv.ParseUint(16).
func parseHex(s []byte) uint {
n := uint(0)
for i := 0; i < len(s); i++ {
n *= 16
n += uint(fromHexChar(s[i]))
}
return n
}
// Simplified copy of strconv.ParseUint(10).
func parseDec(s []byte) uint64 {
n := uint64(0)
for _, c := range s {
n *= 10
n += uint64(c - '0')
}
return n
}
// hexDecode32big decodes sequences of 32bit big endian bytes.
func hexDecode32big(src []byte) []byte {
dst := make([]byte, len(src) / 2)
blocks := len(src) / 8
for block := 0; block < blocks; block++ {
for i := 0; i < 4; i++ {
a := fromHexChar(src[block * 8 + i * 2])
b := fromHexChar(src[block * 8 + i * 2 + 1])
dst[block * 4 + 3 - i] = (a << 4) | b
}
}
return dst
}
// fromHexChar converts a hex character into its value.
func fromHexChar(c byte) uint8 {
switch {
case '0' <= c && c <= '9':
return c - '0'
case 'a' <= c && c <= 'f':
return c - 'a' + 10
case 'A' <= c && c <= 'F':
return c - 'A' + 10
}
return 0
}

40
probe/proc/spy.go Normal file
View File

@@ -0,0 +1,40 @@
package proc
import (
"net"
)
const (
tcpEstablished = 1 // according to /include/net/tcp_states.h
)
// Connection is a (TCP) connection. The Proc struct might not be filled in.
type Connection struct {
Transport string
LocalAddress net.IP
LocalPort uint16
RemoteAddress net.IP
RemotePort uint16
inode uint64
Proc
}
// Proc is a single process with PID and process name.
type Proc struct {
PID uint
Name string
}
// ConnIter is returned by Connections().
type ConnIter interface {
Next() *Connection
}
// Connections returns all established (TCP) connections. If processes is
// false we'll just list all TCP connections, and there is no need to be root.
// If processes is true it'll additionally try to lookup the process owning the
// connection, filling in the Proc field. You will need to run this as root to
// find all processes.
func Connections(processes bool) (ConnIter, error) {
return cbConnections(processes)
}

52
probe/proc/spy_linux.go Normal file
View File

@@ -0,0 +1,52 @@
package proc
import (
"bytes"
"path"
"sync"
)
var bufPool = sync.Pool{
New: func() interface{} {
return bytes.NewBuffer(make([]byte, 0, 5000))
},
}
type pnConnIter struct {
pn *ProcNet
buf *bytes.Buffer
procs map[uint64]Proc
}
func (c *pnConnIter) Next() *Connection {
n := c.pn.Next()
if n == nil {
// Done!
bufPool.Put(c.buf)
return nil
}
if proc, ok := c.procs[n.inode]; ok {
n.Proc = proc
}
return n
}
// cbConnections sets Connections()
var cbConnections = func(processes bool) (ConnIter, error) {
buf := bufPool.Get().(*bytes.Buffer)
buf.Reset()
readFile(path.Join(procRoot, "net", "tcp"), buf)
readFile(path.Join(procRoot, "net", "tcp6"), buf)
var procs map[uint64]Proc
if processes {
var err error
if procs, err = walkProcPid(); err != nil {
return nil, err
}
}
return &pnConnIter{
pn: NewProcNet(buf.Bytes(), tcpEstablished),
buf: buf,
procs: procs,
}, nil
}

View File

@@ -1,4 +1,4 @@
package process
package proc
import (
"fmt"

View File

@@ -1,15 +1,15 @@
package process_test
package proc_test
import (
"reflect"
"testing"
"github.com/weaveworks/scope/probe/process"
"github.com/weaveworks/scope/probe/proc"
)
func TestTree(t *testing.T) {
walker := &mockWalker{
processes: []process.Process{
processes: []proc.Process{
{PID: 1, PPID: 0, Comm: "init"},
{PID: 2, PPID: 1, Comm: "bash"},
{PID: 3, PPID: 1, Comm: "apache", Threads: 2},
@@ -17,7 +17,7 @@ func TestTree(t *testing.T) {
},
}
tree, err := process.NewTree(walker)
tree, err := proc.NewTree(walker)
if err != nil {
t.Fatalf("newProcessTree error: %v", err)
}

View File

@@ -1,4 +1,4 @@
package process
package proc
import "sync"

View File

@@ -1,4 +1,4 @@
package process
package proc
import (
"fmt"

View File

@@ -1,4 +1,4 @@
package process
package proc
import (
"bytes"

View File

@@ -1,4 +1,4 @@
package process_test
package proc_test
import (
"fmt"
@@ -9,7 +9,7 @@ import (
"testing"
"time"
"github.com/weaveworks/scope/probe/process"
"github.com/weaveworks/scope/probe/proc"
"github.com/weaveworks/scope/test"
)
@@ -25,10 +25,10 @@ func (p mockProcess) IsDir() bool { return true }
func (p mockProcess) Sys() interface{} { return nil }
func TestWalker(t *testing.T) {
oldReadDir, oldReadFile := process.ReadDir, process.ReadFile
oldReadDir, oldReadFile := proc.ReadDir, proc.ReadFile
defer func() {
process.ReadDir = oldReadDir
process.ReadFile = oldReadFile
proc.ReadDir = oldReadDir
proc.ReadFile = oldReadFile
}()
processes := map[string]mockProcess{
@@ -39,7 +39,7 @@ func TestWalker(t *testing.T) {
"1": {name: "1", comm: "init\n"},
}
process.ReadDir = func(path string) ([]os.FileInfo, error) {
proc.ReadDir = func(path string) ([]os.FileInfo, error) {
result := []os.FileInfo{}
for _, p := range processes {
result = append(result, p)
@@ -47,7 +47,7 @@ func TestWalker(t *testing.T) {
return result, nil
}
process.ReadFile = func(path string) ([]byte, error) {
proc.ReadFile = func(path string) ([]byte, error) {
splits := strings.Split(path, "/")
pid := splits[len(splits)-2]
@@ -71,16 +71,16 @@ func TestWalker(t *testing.T) {
return nil, fmt.Errorf("not found")
}
want := map[int]process.Process{
want := map[int]proc.Process{
3: {PID: 3, PPID: 2, Comm: "curl", Cmdline: "curl google.com", Threads: 1},
2: {PID: 2, PPID: 1, Comm: "bash", Cmdline: "", Threads: 1},
4: {PID: 4, PPID: 3, Comm: "apache", Cmdline: "", Threads: 1},
1: {PID: 1, PPID: 0, Comm: "init", Cmdline: "", Threads: 1},
}
have := map[int]process.Process{}
walker := process.NewWalker("unused")
err := walker.Walk(func(p process.Process) {
have := map[int]proc.Process{}
walker := proc.NewWalker("unused")
err := walker.Walk(func(p proc.Process) {
have[p.PID] = p
})

View File

@@ -1,25 +1,36 @@
package process_test
package proc_test
import (
"reflect"
"testing"
"github.com/weaveworks/scope/probe/process"
"github.com/weaveworks/scope/probe/proc"
"github.com/weaveworks/scope/test"
)
type mockWalker struct {
processes []proc.Process
}
func (m *mockWalker) Walk(f func(proc.Process)) error {
for _, p := range m.processes {
f(p)
}
return nil
}
func TestBasicWalk(t *testing.T) {
var (
procRoot = "/proc"
procFunc = func(process.Process) {}
procFunc = func(proc.Process) {}
)
if err := process.NewWalker(procRoot).Walk(procFunc); err != nil {
if err := proc.NewWalker(procRoot).Walk(procFunc); err != nil {
t.Fatal(err)
}
}
func TestCache(t *testing.T) {
processes := []process.Process{
processes := []proc.Process{
{PID: 1, PPID: 0, Comm: "init"},
{PID: 2, PPID: 1, Comm: "bash"},
{PID: 3, PPID: 1, Comm: "apache", Threads: 2},
@@ -28,7 +39,7 @@ func TestCache(t *testing.T) {
walker := &mockWalker{
processes: processes,
}
cachingWalker := process.NewCachingWalker(walker)
cachingWalker := proc.NewCachingWalker(walker)
err := cachingWalker.Tick()
if err != nil {
t.Fatal(err)
@@ -39,7 +50,7 @@ func TestCache(t *testing.T) {
t.Errorf("%v (%v)", test.Diff(processes, have), err)
}
walker.processes = []process.Process{}
walker.processes = []proc.Process{}
have, err = all(cachingWalker)
if err != nil || !reflect.DeepEqual(processes, have) {
t.Errorf("%v (%v)", test.Diff(processes, have), err)
@@ -51,15 +62,15 @@ func TestCache(t *testing.T) {
}
have, err = all(cachingWalker)
want := []process.Process{}
want := []proc.Process{}
if err != nil || !reflect.DeepEqual(want, have) {
t.Errorf("%v (%v)", test.Diff(want, have), err)
}
}
func all(w process.Walker) ([]process.Process, error) {
all := []process.Process{}
err := w.Walk(func(p process.Process) {
func all(w proc.Walker) ([]proc.Process, error) {
all := []proc.Process{}
err := w.Walk(func(p proc.Process) {
all = append(all, p)
})
return all, err

View File

@@ -4,6 +4,7 @@ import (
"strconv"
"github.com/weaveworks/scope/report"
"github.com/weaveworks/scope/probe/proc"
)
// We use these keys in node metadata
@@ -18,11 +19,11 @@ const (
// Reporter generates Reports containing the Process topology.
type Reporter struct {
scope string
walker Walker
walker proc.Walker
}
// NewReporter makes a new Reporter.
func NewReporter(walker Walker, scope string) *Reporter {
func NewReporter(walker proc.Walker, scope string) *Reporter {
return &Reporter{
scope: scope,
walker: walker,
@@ -42,7 +43,7 @@ func (r *Reporter) Report() (report.Report, error) {
func (r *Reporter) processTopology() (report.Topology, error) {
t := report.MakeTopology()
err := r.walker.Walk(func(p Process) {
err := r.walker.Walk(func(p proc.Process) {
pidstr := strconv.Itoa(p.PID)
nodeID := report.MakeProcessNodeID(r.scope, pidstr)
t.Nodes[nodeID] = report.MakeNode()

View File

@@ -4,16 +4,17 @@ import (
"reflect"
"testing"
"github.com/weaveworks/scope/probe/proc"
"github.com/weaveworks/scope/probe/process"
"github.com/weaveworks/scope/report"
"github.com/weaveworks/scope/test"
)
type mockWalker struct {
processes []process.Process
processes []proc.Process
}
func (m *mockWalker) Walk(f func(process.Process)) error {
func (m *mockWalker) Walk(f func(proc.Process)) error {
for _, p := range m.processes {
f(p)
}
@@ -22,7 +23,7 @@ func (m *mockWalker) Walk(f func(process.Process)) error {
func TestReporter(t *testing.T) {
walker := &mockWalker{
processes: []process.Process{
processes: []proc.Process{
{PID: 1, PPID: 0, Comm: "init"},
{PID: 2, PPID: 1, Comm: "bash"},
{PID: 3, PPID: 1, Comm: "apache", Threads: 2},