moby/daemon/prune.go
Cory Snider d222bf097c daemon: reload runtimes w/o breaking containers
The existing runtimes reload logic went to great lengths to replace the
directory containing runtime wrapper scripts as atomically as possible
within the limitations of the Linux filesystem ABI. Trouble is,
atomically swapping the wrapper scripts directory solves the wrong
problem! The runtime configuration is "locked in" when a container is
started, including the path to the runC binary. If a container is
started with a runtime which requires a daemon-managed wrapper script
and then the daemon is reloaded with a config which no longer requires
the wrapper script (i.e. some args -> no args, or the runtime is dropped
from the config), that container would become unmanageable. Any attempts
to stop, exec or otherwise perform lifecycle management operations on
the container are likely to fail due to the wrapper script no longer
existing at its original path.

Atomically swapping the wrapper scripts is also incompatible with the
read-copy-update paradigm for reloading configuration. A handler in the
daemon could retain a reference to the pre-reload configuration for an
indeterminate amount of time after the daemon configuration has been
reloaded and updated. It is possible for the daemon to attempt to start
a container using a deleted wrapper script if a request to run a
container races a reload.

Solve the problem of deleting referenced wrapper scripts by ensuring
that all wrapper scripts are *immutable* for the lifetime of the daemon
process. Any given runtime wrapper script must always exist with the
same contents, no matter how many times the daemon config is reloaded,
or what changes are made to the config. This is accomplished by using
everyone's favourite design pattern: content-addressable storage. Each
wrapper script file name is suffixed with the SHA-256 digest of its
contents to (probabilistically) guarantee immutability without needing
any concurrency control. Stale runtime wrapper scripts are only cleaned
up on the next daemon restart.

Split the derived runtimes configuration from the user-supplied
configuration to have a place to store derived state without mutating
the user-supplied configuration or exposing daemon internals in API
struct types. Hold the derived state and the user-supplied configuration
in a single struct value so that they can be updated as an atomic unit.

Signed-off-by: Cory Snider <csnider@mirantis.com>
2023-06-01 14:45:25 -04:00

259 lines
7.1 KiB
Go

package daemon // import "github.com/docker/docker/daemon"
import (
"context"
"regexp"
"strconv"
"sync/atomic"
"time"
"github.com/docker/docker/api/types"
"github.com/docker/docker/api/types/events"
"github.com/docker/docker/api/types/filters"
timetypes "github.com/docker/docker/api/types/time"
"github.com/docker/docker/errdefs"
"github.com/docker/docker/libnetwork"
"github.com/docker/docker/runconfig"
"github.com/pkg/errors"
"github.com/sirupsen/logrus"
)
var (
// errPruneRunning is returned when a prune request is received while
// one is in progress
errPruneRunning = errdefs.Conflict(errors.New("a prune operation is already running"))
containersAcceptedFilters = map[string]bool{
"label": true,
"label!": true,
"until": true,
}
networksAcceptedFilters = map[string]bool{
"label": true,
"label!": true,
"until": true,
}
)
// ContainersPrune removes unused containers
func (daemon *Daemon) ContainersPrune(ctx context.Context, pruneFilters filters.Args) (*types.ContainersPruneReport, error) {
if !atomic.CompareAndSwapInt32(&daemon.pruneRunning, 0, 1) {
return nil, errPruneRunning
}
defer atomic.StoreInt32(&daemon.pruneRunning, 0)
rep := &types.ContainersPruneReport{}
// make sure that only accepted filters have been received
err := pruneFilters.Validate(containersAcceptedFilters)
if err != nil {
return nil, err
}
until, err := getUntilFromPruneFilters(pruneFilters)
if err != nil {
return nil, err
}
cfg := &daemon.config().Config
allContainers := daemon.List()
for _, c := range allContainers {
select {
case <-ctx.Done():
logrus.Debugf("ContainersPrune operation cancelled: %#v", *rep)
return rep, nil
default:
}
if !c.IsRunning() {
if !until.IsZero() && c.Created.After(until) {
continue
}
if !matchLabels(pruneFilters, c.Config.Labels) {
continue
}
cSize, _, err := daemon.imageService.GetContainerLayerSize(ctx, c.ID)
if err != nil {
return nil, err
}
// TODO: sets RmLink to true?
err = daemon.containerRm(cfg, c.ID, &types.ContainerRmConfig{})
if err != nil {
logrus.Warnf("failed to prune container %s: %v", c.ID, err)
continue
}
if cSize > 0 {
rep.SpaceReclaimed += uint64(cSize)
}
rep.ContainersDeleted = append(rep.ContainersDeleted, c.ID)
}
}
daemon.EventsService.Log("prune", events.ContainerEventType, events.Actor{
Attributes: map[string]string{"reclaimed": strconv.FormatUint(rep.SpaceReclaimed, 10)},
})
return rep, nil
}
// localNetworksPrune removes unused local networks
func (daemon *Daemon) localNetworksPrune(ctx context.Context, pruneFilters filters.Args) *types.NetworksPruneReport {
rep := &types.NetworksPruneReport{}
until, _ := getUntilFromPruneFilters(pruneFilters)
// When the function returns true, the walk will stop.
l := func(nw libnetwork.Network) bool {
select {
case <-ctx.Done():
// context cancelled
return true
default:
}
if nw.Info().ConfigOnly() {
return false
}
if !until.IsZero() && nw.Info().Created().After(until) {
return false
}
if !matchLabels(pruneFilters, nw.Info().Labels()) {
return false
}
nwName := nw.Name()
if runconfig.IsPreDefinedNetwork(nwName) {
return false
}
if len(nw.Endpoints()) > 0 {
return false
}
if err := daemon.DeleteNetwork(nw.ID()); err != nil {
logrus.Warnf("could not remove local network %s: %v", nwName, err)
return false
}
rep.NetworksDeleted = append(rep.NetworksDeleted, nwName)
return false
}
daemon.netController.WalkNetworks(l)
return rep
}
// clusterNetworksPrune removes unused cluster networks
func (daemon *Daemon) clusterNetworksPrune(ctx context.Context, pruneFilters filters.Args) (*types.NetworksPruneReport, error) {
rep := &types.NetworksPruneReport{}
until, _ := getUntilFromPruneFilters(pruneFilters)
cluster := daemon.GetCluster()
if !cluster.IsManager() {
return rep, nil
}
networks, err := cluster.GetNetworks(pruneFilters)
if err != nil {
return rep, err
}
networkIsInUse := regexp.MustCompile(`network ([[:alnum:]]+) is in use`)
for _, nw := range networks {
select {
case <-ctx.Done():
return rep, nil
default:
if nw.Ingress {
// Routing-mesh network removal has to be explicitly invoked by user
continue
}
if !until.IsZero() && nw.Created.After(until) {
continue
}
if !matchLabels(pruneFilters, nw.Labels) {
continue
}
// https://github.com/docker/docker/issues/24186
// `docker network inspect` unfortunately displays ONLY those containers that are local to that node.
// So we try to remove it anyway and check the error
err = cluster.RemoveNetwork(nw.ID)
if err != nil {
// we can safely ignore the "network .. is in use" error
match := networkIsInUse.FindStringSubmatch(err.Error())
if len(match) != 2 || match[1] != nw.ID {
logrus.Warnf("could not remove cluster network %s: %v", nw.Name, err)
}
continue
}
rep.NetworksDeleted = append(rep.NetworksDeleted, nw.Name)
}
}
return rep, nil
}
// NetworksPrune removes unused networks
func (daemon *Daemon) NetworksPrune(ctx context.Context, pruneFilters filters.Args) (*types.NetworksPruneReport, error) {
if !atomic.CompareAndSwapInt32(&daemon.pruneRunning, 0, 1) {
return nil, errPruneRunning
}
defer atomic.StoreInt32(&daemon.pruneRunning, 0)
// make sure that only accepted filters have been received
err := pruneFilters.Validate(networksAcceptedFilters)
if err != nil {
return nil, err
}
if _, err := getUntilFromPruneFilters(pruneFilters); err != nil {
return nil, err
}
rep := &types.NetworksPruneReport{}
if clusterRep, err := daemon.clusterNetworksPrune(ctx, pruneFilters); err == nil {
rep.NetworksDeleted = append(rep.NetworksDeleted, clusterRep.NetworksDeleted...)
}
localRep := daemon.localNetworksPrune(ctx, pruneFilters)
rep.NetworksDeleted = append(rep.NetworksDeleted, localRep.NetworksDeleted...)
select {
case <-ctx.Done():
logrus.Debugf("NetworksPrune operation cancelled: %#v", *rep)
return rep, nil
default:
}
daemon.EventsService.Log("prune", events.NetworkEventType, events.Actor{
Attributes: map[string]string{"reclaimed": "0"},
})
return rep, nil
}
func getUntilFromPruneFilters(pruneFilters filters.Args) (time.Time, error) {
until := time.Time{}
if !pruneFilters.Contains("until") {
return until, nil
}
untilFilters := pruneFilters.Get("until")
if len(untilFilters) > 1 {
return until, errdefs.InvalidParameter(errors.New("more than one until filter specified"))
}
ts, err := timetypes.GetTimestamp(untilFilters[0], time.Now())
if err != nil {
return until, errdefs.InvalidParameter(err)
}
seconds, nanoseconds, err := timetypes.ParseTimestamps(ts, 0)
if err != nil {
return until, errdefs.InvalidParameter(err)
}
until = time.Unix(seconds, nanoseconds)
return until, nil
}
func matchLabels(pruneFilters filters.Args, labels map[string]string) bool {
if !pruneFilters.MatchKVList("label", labels) {
return false
}
// By default MatchKVList will return true if field (like 'label!') does not exist
// So we have to add additional Contains("label!") check
if pruneFilters.Contains("label!") {
if pruneFilters.MatchKVList("label!", labels) {
return false
}
}
return true
}