gotosocial/vendor/github.com/containerd/cgroups/blkio.go
Daniele Sluijters acc333c40b
[feature] Inherit resource limits from cgroups (#1336)
When GTS is running in a container runtime which has configured CPU or
memory limits or under an init system that uses cgroups to impose CPU
and memory limits the values the Go runtime sees for GOMAXPROCS and
GOMEMLIMIT are still based on the host resources, not the cgroup.

At least for the throttling middlewares which use GOMAXPROCS to
configure their queue size, this can result in GTS running with values
too big compared to the resources that will actuall be available to it.

This introduces 2 dependencies which can pick up resource contraints
from the current cgroup and tune the Go runtime accordingly. This should
result in the different queues being appropriately sized and in general
more predictable performance. These dependencies are a no-op on
non-Linux systems or if running in a cgroup that doesn't set a limit on
CPU or memory.

The automatic tuning of GOMEMLIMIT can be disabled by either explicitly
setting GOMEMLIMIT yourself or by setting AUTOMEMLIMIT=off. The
automatic tuning of GOMAXPROCS can similarly be counteracted by setting
GOMAXPROCS yourself.
2023-01-17 20:59:04 +00:00

362 lines
8.1 KiB
Go

/*
Copyright The containerd Authors.
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
*/
package cgroups
import (
"bufio"
"fmt"
"io"
"os"
"path/filepath"
"strconv"
"strings"
v1 "github.com/containerd/cgroups/stats/v1"
specs "github.com/opencontainers/runtime-spec/specs-go"
)
// NewBlkio returns a Blkio controller given the root folder of cgroups.
// It may optionally accept other configuration options, such as ProcRoot(path)
func NewBlkio(root string, options ...func(controller *blkioController)) *blkioController {
ctrl := &blkioController{
root: filepath.Join(root, string(Blkio)),
procRoot: "/proc",
}
for _, opt := range options {
opt(ctrl)
}
return ctrl
}
// ProcRoot overrides the default location of the "/proc" filesystem
func ProcRoot(path string) func(controller *blkioController) {
return func(c *blkioController) {
c.procRoot = path
}
}
type blkioController struct {
root string
procRoot string
}
func (b *blkioController) Name() Name {
return Blkio
}
func (b *blkioController) Path(path string) string {
return filepath.Join(b.root, path)
}
func (b *blkioController) Create(path string, resources *specs.LinuxResources) error {
if err := os.MkdirAll(b.Path(path), defaultDirPerm); err != nil {
return err
}
if resources.BlockIO == nil {
return nil
}
for _, t := range createBlkioSettings(resources.BlockIO) {
if t.value != nil {
if err := retryingWriteFile(
filepath.Join(b.Path(path), "blkio."+t.name),
t.format(t.value),
defaultFilePerm,
); err != nil {
return err
}
}
}
return nil
}
func (b *blkioController) Update(path string, resources *specs.LinuxResources) error {
return b.Create(path, resources)
}
func (b *blkioController) Stat(path string, stats *v1.Metrics) error {
stats.Blkio = &v1.BlkIOStat{}
var settings []blkioStatSettings
// Try to read CFQ stats available on all CFQ enabled kernels first
if _, err := os.Lstat(filepath.Join(b.Path(path), "blkio.io_serviced_recursive")); err == nil {
settings = []blkioStatSettings{
{
name: "sectors_recursive",
entry: &stats.Blkio.SectorsRecursive,
},
{
name: "io_service_bytes_recursive",
entry: &stats.Blkio.IoServiceBytesRecursive,
},
{
name: "io_serviced_recursive",
entry: &stats.Blkio.IoServicedRecursive,
},
{
name: "io_queued_recursive",
entry: &stats.Blkio.IoQueuedRecursive,
},
{
name: "io_service_time_recursive",
entry: &stats.Blkio.IoServiceTimeRecursive,
},
{
name: "io_wait_time_recursive",
entry: &stats.Blkio.IoWaitTimeRecursive,
},
{
name: "io_merged_recursive",
entry: &stats.Blkio.IoMergedRecursive,
},
{
name: "time_recursive",
entry: &stats.Blkio.IoTimeRecursive,
},
}
}
f, err := os.Open(filepath.Join(b.procRoot, "partitions"))
if err != nil {
return err
}
defer f.Close()
devices, err := getDevices(f)
if err != nil {
return err
}
var size int
for _, t := range settings {
if err := b.readEntry(devices, path, t.name, t.entry); err != nil {
return err
}
size += len(*t.entry)
}
if size > 0 {
return nil
}
// Even the kernel is compiled with the CFQ scheduler, the cgroup may not use
// block devices with the CFQ scheduler. If so, we should fallback to throttle.* files.
settings = []blkioStatSettings{
{
name: "throttle.io_serviced",
entry: &stats.Blkio.IoServicedRecursive,
},
{
name: "throttle.io_service_bytes",
entry: &stats.Blkio.IoServiceBytesRecursive,
},
}
for _, t := range settings {
if err := b.readEntry(devices, path, t.name, t.entry); err != nil {
return err
}
}
return nil
}
func (b *blkioController) readEntry(devices map[deviceKey]string, path, name string, entry *[]*v1.BlkIOEntry) error {
f, err := os.Open(filepath.Join(b.Path(path), "blkio."+name))
if err != nil {
return err
}
defer f.Close()
sc := bufio.NewScanner(f)
for sc.Scan() {
// format: dev type amount
fields := strings.FieldsFunc(sc.Text(), splitBlkIOStatLine)
if len(fields) < 3 {
if len(fields) == 2 && fields[0] == "Total" {
// skip total line
continue
} else {
return fmt.Errorf("invalid line found while parsing %s: %s", path, sc.Text())
}
}
major, err := strconv.ParseUint(fields[0], 10, 64)
if err != nil {
return err
}
minor, err := strconv.ParseUint(fields[1], 10, 64)
if err != nil {
return err
}
op := ""
valueField := 2
if len(fields) == 4 {
op = fields[2]
valueField = 3
}
v, err := strconv.ParseUint(fields[valueField], 10, 64)
if err != nil {
return err
}
*entry = append(*entry, &v1.BlkIOEntry{
Device: devices[deviceKey{major, minor}],
Major: major,
Minor: minor,
Op: op,
Value: v,
})
}
return sc.Err()
}
func createBlkioSettings(blkio *specs.LinuxBlockIO) []blkioSettings {
settings := []blkioSettings{}
if blkio.Weight != nil {
settings = append(settings,
blkioSettings{
name: "weight",
value: blkio.Weight,
format: uintf,
})
}
if blkio.LeafWeight != nil {
settings = append(settings,
blkioSettings{
name: "leaf_weight",
value: blkio.LeafWeight,
format: uintf,
})
}
for _, wd := range blkio.WeightDevice {
if wd.Weight != nil {
settings = append(settings,
blkioSettings{
name: "weight_device",
value: wd,
format: weightdev,
})
}
if wd.LeafWeight != nil {
settings = append(settings,
blkioSettings{
name: "leaf_weight_device",
value: wd,
format: weightleafdev,
})
}
}
for _, t := range []struct {
name string
list []specs.LinuxThrottleDevice
}{
{
name: "throttle.read_bps_device",
list: blkio.ThrottleReadBpsDevice,
},
{
name: "throttle.read_iops_device",
list: blkio.ThrottleReadIOPSDevice,
},
{
name: "throttle.write_bps_device",
list: blkio.ThrottleWriteBpsDevice,
},
{
name: "throttle.write_iops_device",
list: blkio.ThrottleWriteIOPSDevice,
},
} {
for _, td := range t.list {
settings = append(settings, blkioSettings{
name: t.name,
value: td,
format: throttleddev,
})
}
}
return settings
}
type blkioSettings struct {
name string
value interface{}
format func(v interface{}) []byte
}
type blkioStatSettings struct {
name string
entry *[]*v1.BlkIOEntry
}
func uintf(v interface{}) []byte {
return []byte(strconv.FormatUint(uint64(*v.(*uint16)), 10))
}
func weightdev(v interface{}) []byte {
wd := v.(specs.LinuxWeightDevice)
return []byte(fmt.Sprintf("%d:%d %d", wd.Major, wd.Minor, *wd.Weight))
}
func weightleafdev(v interface{}) []byte {
wd := v.(specs.LinuxWeightDevice)
return []byte(fmt.Sprintf("%d:%d %d", wd.Major, wd.Minor, *wd.LeafWeight))
}
func throttleddev(v interface{}) []byte {
td := v.(specs.LinuxThrottleDevice)
return []byte(fmt.Sprintf("%d:%d %d", td.Major, td.Minor, td.Rate))
}
func splitBlkIOStatLine(r rune) bool {
return r == ' ' || r == ':'
}
type deviceKey struct {
major, minor uint64
}
// getDevices makes a best effort attempt to read all the devices into a map
// keyed by major and minor number. Since devices may be mapped multiple times,
// we err on taking the first occurrence.
func getDevices(r io.Reader) (map[deviceKey]string, error) {
var (
s = bufio.NewScanner(r)
devices = make(map[deviceKey]string)
)
for i := 0; s.Scan(); i++ {
if i < 2 {
continue
}
fields := strings.Fields(s.Text())
major, err := strconv.Atoi(fields[0])
if err != nil {
return nil, err
}
minor, err := strconv.Atoi(fields[1])
if err != nil {
return nil, err
}
key := deviceKey{
major: uint64(major),
minor: uint64(minor),
}
if _, ok := devices[key]; ok {
continue
}
devices[key] = filepath.Join("/dev", fields[3])
}
return devices, s.Err()
}