2e5c6de820
This PR introduces support for using Nomad on systems with cgroups v2 [1] enabled as the cgroups controller mounted on /sys/fs/cgroups. Newer Linux distros like Ubuntu 21.10 are shipping with cgroups v2 only, causing problems for Nomad users. Nomad mostly "just works" with cgroups v2 due to the indirection via libcontainer, but not so for managing cpuset cgroups. Before, Nomad has been making use of a feature in v1 where a PID could be a member of more than one cgroup. In v2 this is no longer possible, and so the logic around computing cpuset values must be modified. When Nomad detects v2, it manages cpuset values in-process, rather than making use of cgroup heirarchy inheritence via shared/reserved parents. Nomad will only activate the v2 logic when it detects cgroups2 is mounted at /sys/fs/cgroups. This means on systems running in hybrid mode with cgroups2 mounted at /sys/fs/cgroups/unified (as is typical) Nomad will continue to use the v1 logic, and should operate as before. Systems that do not support cgroups v2 are also not affected. When v2 is activated, Nomad will create a parent called nomad.slice (unless otherwise configured in Client conifg), and create cgroups for tasks using naming convention <allocID>-<task>.scope. These follow the naming convention set by systemd and also used by Docker when cgroups v2 is detected. Client nodes now export a new fingerprint attribute, unique.cgroups.version which will be set to 'v1' or 'v2' to indicate the cgroups regime in use by Nomad. The new cpuset management strategy fixes #11705, where docker tasks that spawned processes on startup would "leak". In cgroups v2, the PIDs are started in the cgroup they will always live in, and thus the cause of the leak is eliminated. [1] https://www.kernel.org/doc/html/latest/admin-guide/cgroup-v2.html Closes #11289 Fixes #11705 #11773 #11933
126 lines
2.8 KiB
Go
126 lines
2.8 KiB
Go
//go:build linux
|
|
|
|
package docker
|
|
|
|
import (
|
|
"context"
|
|
"fmt"
|
|
"path/filepath"
|
|
"sync"
|
|
"time"
|
|
|
|
"github.com/hashicorp/go-hclog"
|
|
"github.com/hashicorp/nomad/client/lib/cgutil"
|
|
"github.com/hashicorp/nomad/helper"
|
|
)
|
|
|
|
const (
|
|
cpusetReconcileInterval = 1 * time.Second
|
|
)
|
|
|
|
type CpusetFixer interface {
|
|
Start()
|
|
}
|
|
|
|
// cpusetFixer adjusts the cpuset.cpus cgroup value to the assigned value by Nomad.
|
|
//
|
|
// Due to Docker not allowing the configuration of the full cgroup path, we must
|
|
// manually fix the cpuset values for all docker containers continuously, as the
|
|
// values will change as tasks of any driver using reserved cores are started and
|
|
// stopped, changing the size of the remaining shared cpu pool.
|
|
//
|
|
// The exec/java, podman, and containerd runtimes let you specify the cgroup path,
|
|
// making use of the cgroup Nomad creates and manages on behalf of the task.
|
|
type cpusetFixer struct {
|
|
ctx context.Context
|
|
logger hclog.Logger
|
|
interval time.Duration
|
|
once sync.Once
|
|
tasks func() map[coordinate]struct{}
|
|
}
|
|
|
|
func newCpusetFixer(d *Driver) CpusetFixer {
|
|
return &cpusetFixer{
|
|
interval: cpusetReconcileInterval,
|
|
ctx: d.ctx,
|
|
logger: d.logger,
|
|
tasks: d.trackedTasks,
|
|
}
|
|
}
|
|
|
|
// Start will start the background cpuset reconciliation until the cf context is
|
|
// cancelled for shutdown.
|
|
//
|
|
// Only runs if cgroups.v2 is in use.
|
|
func (cf *cpusetFixer) Start() {
|
|
cf.once.Do(func() {
|
|
if cgutil.UseV2 {
|
|
go cf.loop()
|
|
}
|
|
})
|
|
}
|
|
|
|
func (cf *cpusetFixer) loop() {
|
|
timer, cancel := helper.NewSafeTimer(0)
|
|
defer cancel()
|
|
|
|
for {
|
|
select {
|
|
case <-cf.ctx.Done():
|
|
return
|
|
case <-timer.C:
|
|
timer.Stop()
|
|
cf.apply()
|
|
timer.Reset(cf.interval)
|
|
}
|
|
}
|
|
}
|
|
|
|
func (cf *cpusetFixer) apply() {
|
|
coordinates := cf.tasks()
|
|
for c := range coordinates {
|
|
cf.fix(c)
|
|
}
|
|
}
|
|
|
|
func (cf *cpusetFixer) fix(c coordinate) {
|
|
source := c.NomadCgroup()
|
|
destination := c.DockerCgroup()
|
|
if err := cgutil.CopyCpuset(source, destination); err != nil {
|
|
cf.logger.Debug("failed to copy cpuset", "err", err)
|
|
}
|
|
}
|
|
|
|
type coordinate struct {
|
|
containerID string
|
|
allocID string
|
|
task string
|
|
path string
|
|
}
|
|
|
|
func (c coordinate) NomadCgroup() string {
|
|
parent, _ := cgutil.SplitPath(c.path)
|
|
return filepath.Join(cgutil.CgroupRoot, parent, cgutil.CgroupScope(c.allocID, c.task))
|
|
}
|
|
|
|
func (c coordinate) DockerCgroup() string {
|
|
parent, _ := cgutil.SplitPath(c.path)
|
|
return filepath.Join(cgutil.CgroupRoot, parent, fmt.Sprintf("docker-%s.scope", c.containerID))
|
|
}
|
|
|
|
func (d *Driver) trackedTasks() map[coordinate]struct{} {
|
|
d.tasks.lock.RLock()
|
|
defer d.tasks.lock.RUnlock()
|
|
|
|
m := make(map[coordinate]struct{}, len(d.tasks.store))
|
|
for _, h := range d.tasks.store {
|
|
m[coordinate{
|
|
containerID: h.containerID,
|
|
allocID: h.task.AllocID,
|
|
task: h.task.Name,
|
|
path: h.task.Resources.LinuxResources.CpusetCgroupPath,
|
|
}] = struct{}{}
|
|
}
|
|
return m
|
|
}
|