1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270
|
//go:build !remote
package libpod
import (
"crypto/rand"
jdec "encoding/json"
"errors"
"fmt"
"net"
"os/exec"
"path/filepath"
"github.com/containers/buildah/pkg/jail"
"github.com/containers/common/libnetwork/types"
"github.com/containers/podman/v5/libpod/define"
"github.com/containers/storage/pkg/lockfile"
"github.com/sirupsen/logrus"
)
type Netstat struct {
Statistics NetstatInterface `json:"statistics"`
}
type NetstatInterface struct {
Interface []NetstatAddress `json:"interface"`
}
type NetstatAddress struct {
Name string `json:"name"`
Flags string `json:"flags"`
Mtu int `json:"mtu"`
Network string `json:"network"`
Address string `json:"address"`
ReceivedPackets uint64 `json:"received-packets"`
ReceivedBytes uint64 `json:"received-bytes"`
ReceivedErrors uint64 `json:"received-errors"`
SentPackets uint64 `json:"sent-packets"`
SentBytes uint64 `json:"sent-bytes"`
SentErrors uint64 `json:"send-errors"`
DroppedPackets uint64 `json:"dropped-packets"`
Collisions uint64 `json:"collisions"`
}
type RootlessNetNS struct {
dir string
Lock *lockfile.LockFile
}
// getPath will join the given path to the rootless netns dir
func (r *RootlessNetNS) getPath(path string) string {
return filepath.Join(r.dir, path)
}
// Do - run the given function in the rootless netns.
// It does not lock the rootlessCNI lock, the caller
// should only lock when needed, e.g. for network operations.
func (r *RootlessNetNS) Do(toRun func() error) error {
return errors.New("not supported on freebsd")
}
// Cleanup the rootless network namespace if needed.
// It checks if we have running containers with the bridge network mode.
// Cleanup() expects that r.Lock is locked
func (r *RootlessNetNS) Cleanup(runtime *Runtime) error {
return errors.New("not supported on freebsd")
}
// GetRootlessNetNs returns the rootless netns object. If create is set to true
// the rootless network namespace will be created if it does not already exist.
// If called as root it returns always nil.
// On success the returned RootlessCNI lock is locked and must be unlocked by the caller.
func (r *Runtime) GetRootlessNetNs(new bool) (*RootlessNetNS, error) {
return nil, nil
}
func getSlirp4netnsIP(subnet *net.IPNet) (*net.IP, error) {
return nil, errors.New("not implemented GetSlirp4netnsIP")
}
// This is called after the container's jail is created but before its
// started. We can use this to initialise the container's vnet when we don't
// have a separate vnet jail (which is the case in FreeBSD 13.3 and later).
func (r *Runtime) setupNetNS(ctr *Container) error {
networkStatus, err := r.configureNetNS(ctr, ctr.ID())
ctr.state.NetNS = ctr.ID()
ctr.state.NetworkStatus = networkStatus
return err
}
// Create and configure a new network namespace for a container
func (r *Runtime) configureNetNS(ctr *Container, ctrNS string) (status map[string]types.StatusBlock, rerr error) {
if err := r.exposeMachinePorts(ctr.config.PortMappings); err != nil {
return nil, err
}
defer func() {
// make sure to unexpose the gvproxy ports when an error happens
if rerr != nil {
if err := r.unexposeMachinePorts(ctr.config.PortMappings); err != nil {
logrus.Errorf("failed to free gvproxy machine ports: %v", err)
}
}
}()
networks, err := ctr.networks()
if err != nil {
return nil, err
}
// All networks have been removed from the container.
// This is effectively forcing net=none.
if len(networks) == 0 {
return nil, nil
}
netOpts := ctr.getNetworkOptions(networks)
netStatus, err := r.setUpNetwork(ctrNS, netOpts)
if err != nil {
return nil, err
}
return netStatus, err
}
// Create and configure a new network namespace for a container
func (r *Runtime) createNetNS(ctr *Container) (n string, q map[string]types.StatusBlock, retErr error) {
b := make([]byte, 16)
_, err := rand.Reader.Read(b)
if err != nil {
return "", nil, fmt.Errorf("failed to generate random vnet name: %v", err)
}
netns := fmt.Sprintf("vnet-%x-%x-%x-%x-%x", b[0:4], b[4:6], b[6:8], b[8:10], b[10:])
jconf := jail.NewConfig()
jconf.Set("name", netns)
jconf.Set("vnet", jail.NEW)
jconf.Set("children.max", 1)
jconf.Set("persist", true)
jconf.Set("enforce_statfs", 0)
jconf.Set("devfs_ruleset", 4)
jconf.Set("allow.raw_sockets", true)
jconf.Set("allow.chflags", true)
jconf.Set("securelevel", -1)
j, err := jail.Create(jconf)
if err != nil {
return "", nil, fmt.Errorf("Failed to create vnet jail %s for container %s: %w", netns, ctr.ID(), err)
}
logrus.Debugf("Created vnet jail %s for container %s", netns, ctr.ID())
var networkStatus map[string]types.StatusBlock
networkStatus, err = r.configureNetNS(ctr, netns)
if err != nil {
jconf := jail.NewConfig()
jconf.Set("persist", false)
if err := j.Set(jconf); err != nil {
// Log this error and return the error from configureNetNS
logrus.Errorf("failed to destroy vnet jail %s: %w", netns, err)
}
}
return netns, networkStatus, err
}
// Tear down a network namespace, undoing all state associated with it.
func (r *Runtime) teardownNetNS(ctr *Container) error {
if err := r.unexposeMachinePorts(ctr.config.PortMappings); err != nil {
// do not return an error otherwise we would prevent network cleanup
logrus.Errorf("failed to free gvproxy machine ports: %v", err)
}
if err := r.teardownNetwork(ctr); err != nil {
return err
}
if ctr.state.NetNS != "" {
// If PostConfigureNetNS is false, then we are running with a
// separate vnet jail so we need to clean that up now.
if !ctr.config.PostConfigureNetNS {
// Rather than destroying the jail immediately, reset the
// persist flag so that it will live until the container is
// done.
netjail, err := jail.FindByName(ctr.state.NetNS)
if err != nil {
return fmt.Errorf("finding network jail %s: %w", ctr.state.NetNS, err)
}
jconf := jail.NewConfig()
jconf.Set("persist", false)
if err := netjail.Set(jconf); err != nil {
return fmt.Errorf("releasing network jail %s: %w", ctr.state.NetNS, err)
}
}
ctr.state.NetNS = ""
}
return nil
}
// TODO (5.0): return the statistics per network interface
// This would allow better compat with docker.
func getContainerNetIO(ctr *Container) (map[string]define.ContainerNetworkStats, error) {
if ctr.state.NetNS == "" {
// If NetNS is nil, it was set as none, and no netNS
// was set up this is a valid state and thus return no
// error, nor any statistics
return nil, nil
}
// First try running 'netstat -j' - this lets us retrieve stats from
// containers which don't have a separate vnet jail.
cmd := exec.Command("netstat", "-j", ctr.state.NetNS, "-bi", "--libxo", "json")
out, err := cmd.Output()
if err != nil {
// Fall back to using jexec so that this still works on 13.2
// which does not have the -j flag.
cmd := exec.Command("jexec", ctr.state.NetNS, "netstat", "-bi", "--libxo", "json")
out, err = cmd.Output()
}
if err != nil {
return nil, fmt.Errorf("failed to read network stats: %v", err)
}
stats := Netstat{}
if err := jdec.Unmarshal(out, &stats); err != nil {
return nil, err
}
res := make(map[string]define.ContainerNetworkStats)
// Sum all the interface stats - in practice only Tx/TxBytes are needed
for _, ifaddr := range stats.Statistics.Interface {
// Each interface has two records, one for link-layer which has
// an MTU field and one for IP which doesn't. We only want the
// link-layer stats.
//
// It's not clear if we should include loopback stats here but
// if we move to per-interface stats in future, this can be
// reported separately.
if ifaddr.Mtu > 0 {
linkStats := define.ContainerNetworkStats{
RxPackets: ifaddr.ReceivedPackets,
TxPackets: ifaddr.SentPackets,
RxBytes: ifaddr.ReceivedBytes,
TxBytes: ifaddr.SentBytes,
RxErrors: ifaddr.ReceivedErrors,
TxErrors: ifaddr.SentErrors,
RxDropped: ifaddr.DroppedPackets,
}
res[ifaddr.Name] = linkStats
}
}
return res, nil
}
func (c *Container) joinedNetworkNSPath() (string, bool) {
return c.state.NetNS, false
}
func (c *Container) inspectJoinedNetworkNS(networkns string) (q types.StatusBlock, retErr error) {
// TODO: extract interface information from the vnet jail
return types.StatusBlock{}, nil
}
func (c *Container) reloadRootlessRLKPortMapping() error {
return errors.New("unsupported (*Container).reloadRootlessRLKPortMapping")
}
func (c *Container) setupRootlessNetwork() error {
return nil
}
|