containerd: add option to set parent cgroup

Using the "runc.v2" runtime, it is possible to configure containerd to start
runc with the "systemd_cgroup" flag. This will cause runc to use systemd to
manage the container cgroups. For this configuration to work, runc needs the
cgroup name to be of a special form: "<systemd.slice>:<parent>:<name>".
This is already implemented in the containerd runtime package, provided that
a parent cgroup of the form "<systemd.slice>:<parent>:" is set.

This commit adds the option to configure such a parent cgroup for the
containerd worker. By default, it will still use an empty string as cgroup
parent, keeping existing behaviour.

Using a configuration like:

  [worker.containerd]
  defaultCgroupParent = "system.slice:buildkit:"
  [worker.containerd.runtime]
    name = "io.containerd.runc.v2"
    [worker.containerd.runtime.options]
      SystemdCgroup = true

a user is able to have their container cgroups managed by systemd. This makes
it possible to set global resource constraints on a per-container basis using
systemd drop-in configuration. When using the example above, the following file
restricts every container spawned by buildkit to use only 1 CPU and 1G of RAM:

  $ cat /etc/systemd/system/buildkit-.scope.d/limits.conf
  [Scope]
  CPUQuota=100%
  MemoryMax=1G

Signed-off-by: Moritz "WanzenBug" Wanzenböck <moritz@wanzenbug.xyz>

cmd/buildkitd/config/config.go

@@ -142,6 +142,8 @@ type ContainerdConfig struct {
 
 	MaxParallelism int `toml:"max-parallelism"`
 
+	DefaultCgroupParent string `toml:"defaultCgroupParent"`
+
 	Rootless bool `toml:"rootless"`
 }
 

cmd/buildkitd/main_containerd_worker.go

@@ -328,7 +328,7 @@ func containerdWorkerInitializer(c *cli.Context, common workerInitializerOpt) ([
 			Options: opts,
 		}
 	}
-	opt, err := containerd.NewWorkerOpt(common.config.Root, cfg.Address, snapshotter, cfg.Namespace, cfg.Rootless, cfg.Labels, dns, nc, common.config.Workers.Containerd.ApparmorProfile, common.config.Workers.Containerd.SELinux, parallelismSem, common.traceSocket, runtime, ctd.WithTimeout(60*time.Second))
+	opt, err := containerd.NewWorkerOpt(common.config.Root, cfg.Address, snapshotter, cfg.Namespace, cfg.DefaultCgroupParent, cfg.Rootless, cfg.Labels, dns, nc, common.config.Workers.Containerd.ApparmorProfile, common.config.Workers.Containerd.SELinux, parallelismSem, common.traceSocket, runtime, ctd.WithTimeout(60*time.Second))
 	if err != nil {
 		return nil, err
 	}

docs/buildkitd.toml.md

@@ -104,6 +104,8 @@ insecure-entitlements = [ "network.host", "security.insecure" ]
   # maintain a pool of reusable CNI network namespaces to amortize the overhead
   # of allocating and releasing the namespaces
   cniPoolSize = 16
+  # defaultCgroupParent sets the parent cgroup of all containers.
+  defaultCgroupParent = "buildkit"
 
   [worker.containerd.labels]
     "foo" = "bar"

worker/containerd/containerd.go

@@ -33,7 +33,7 @@ type RuntimeInfo = containerdexecutor.RuntimeInfo
 // NewWorkerOpt creates a WorkerOpt.
 func NewWorkerOpt(
 	root string,
-	address, snapshotterName, ns string,
+	address, snapshotterName, ns, cgroupParent string,
 	rootless bool,
 	labels map[string]string,
 	dns *oci.DNSConfig,
@@ -62,6 +62,7 @@ func NewWorkerOpt(
 		client,
 		snapshotterName,
 		ns,
+		cgroupParent,
 		rootless,
 		labels,
 		dns,
@@ -74,7 +75,7 @@ func NewWorkerOpt(
 	)
 }
 
-func newContainerd(root string, client *containerd.Client, snapshotterName, ns string, rootless bool, labels map[string]string, dns *oci.DNSConfig, nopt netproviders.Opt, apparmorProfile string, selinux bool, parallelismSem *semaphore.Weighted, traceSocket string, runtime *RuntimeInfo) (base.WorkerOpt, error) {
+func newContainerd(root string, client *containerd.Client, snapshotterName, ns, cgroupParent string, rootless bool, labels map[string]string, dns *oci.DNSConfig, nopt netproviders.Opt, apparmorProfile string, selinux bool, parallelismSem *semaphore.Weighted, traceSocket string, runtime *RuntimeInfo) (base.WorkerOpt, error) {
 	if strings.Contains(snapshotterName, "/") {
 		return base.WorkerOpt{}, errors.Errorf("bad snapshotter name: %q", snapshotterName)
 	}
@@ -176,7 +177,7 @@ func newContainerd(root string, client *containerd.Client, snapshotterName, ns s
 		Labels:           xlabels,
 		MetadataStore:    md,
 		NetworkProviders: np,
-		Executor:         containerdexecutor.New(client, root, "", np, dns, apparmorProfile, selinux, traceSocket, rootless, runtime),
+		Executor:         containerdexecutor.New(client, root, cgroupParent, np, dns, apparmorProfile, selinux, traceSocket, rootless, runtime),
 		Snapshotter:      snap,
 		ContentStore:     cs,
 		Applier:          winlayers.NewFileSystemApplierWithWindows(cs, df),

worker/containerd/containerd_test.go

@@ -31,7 +31,7 @@ func TestContainerdWorkerIntegration(t *testing.T) {
 func newWorkerOpt(t *testing.T, addr string) base.WorkerOpt {
 	tmpdir := t.TempDir()
 	rootless := false
-	workerOpt, err := NewWorkerOpt(tmpdir, addr, "overlayfs", "buildkit-test", rootless, nil, nil, netproviders.Opt{Mode: "host"}, "", false, nil, "", nil)
+	workerOpt, err := NewWorkerOpt(tmpdir, addr, "overlayfs", "buildkit-test", "", rootless, nil, nil, netproviders.Opt{Mode: "host"}, "", false, nil, "", nil)
 	require.NoError(t, err)
 	return workerOpt
 }