moby/daemon/stats.go

144 lines
3.6 KiB
Go
Raw Normal View History

package daemon // import "github.com/docker/docker/daemon"
import (
"context"
"encoding/json"
"errors"
"runtime"
"time"
"github.com/containerd/log"
"github.com/docker/docker/api/types"
"github.com/docker/docker/api/types/backend"
"github.com/docker/docker/container"
Add stats options to not prime the stats Metrics collectors generally don't need the daemon to prime the stats with something to compare since they already have something to compare with. Before this change, the API does 2 collection cycles (which takes roughly 2s) in order to provide comparison for CPU usage over 1s. This was primarily added so that `docker stats --no-stream` had something to compare against. Really the CLI should have just made a 2nd call and done the comparison itself rather than forcing it on all API consumers. That ship has long sailed, though. With this change, clients can set an option to just pull a single stat, which is *at least* a full second faster: Old: ``` time curl --unix-socket /go/src/github.com/docker/docker/bundles/test-integration-shell/docker.sock http://./containers/test/stats?stream=false\&one-shot=false > /dev/null 2>&1 real0m1.864s user0m0.005s sys0m0.007s time curl --unix-socket /go/src/github.com/docker/docker/bundles/test-integration-shell/docker.sock http://./containers/test/stats?stream=false\&one-shot=false > /dev/null 2>&1 real0m1.173s user0m0.010s sys0m0.006s ``` New: ``` time curl --unix-socket /go/src/github.com/docker/docker/bundles/test-integration-shell/docker.sock http://./containers/test/stats?stream=false\&one-shot=true > /dev/null 2>&1 real0m0.680s user0m0.008s sys0m0.004s time curl --unix-socket /go/src/github.com/docker/docker/bundles/test-integration-shell/docker.sock http://./containers/test/stats?stream=false\&one-shot=true > /dev/null 2>&1 real0m0.156s user0m0.007s sys0m0.007s ``` This fixes issues with downstreams ability to use the stats API to collect metrics. Signed-off-by: Brian Goff <cpuguy83@gmail.com>
2020-02-07 23:55:06 +00:00
"github.com/docker/docker/errdefs"
"github.com/docker/docker/pkg/ioutils"
)
// ContainerStats writes information about the container to the stream
// given in the config object.
func (daemon *Daemon) ContainerStats(ctx context.Context, prefixOrName string, config *backend.ContainerStatsConfig) error {
ctr, err := daemon.GetContainer(prefixOrName)
if err != nil {
return err
}
Add stats options to not prime the stats Metrics collectors generally don't need the daemon to prime the stats with something to compare since they already have something to compare with. Before this change, the API does 2 collection cycles (which takes roughly 2s) in order to provide comparison for CPU usage over 1s. This was primarily added so that `docker stats --no-stream` had something to compare against. Really the CLI should have just made a 2nd call and done the comparison itself rather than forcing it on all API consumers. That ship has long sailed, though. With this change, clients can set an option to just pull a single stat, which is *at least* a full second faster: Old: ``` time curl --unix-socket /go/src/github.com/docker/docker/bundles/test-integration-shell/docker.sock http://./containers/test/stats?stream=false\&one-shot=false > /dev/null 2>&1 real0m1.864s user0m0.005s sys0m0.007s time curl --unix-socket /go/src/github.com/docker/docker/bundles/test-integration-shell/docker.sock http://./containers/test/stats?stream=false\&one-shot=false > /dev/null 2>&1 real0m1.173s user0m0.010s sys0m0.006s ``` New: ``` time curl --unix-socket /go/src/github.com/docker/docker/bundles/test-integration-shell/docker.sock http://./containers/test/stats?stream=false\&one-shot=true > /dev/null 2>&1 real0m0.680s user0m0.008s sys0m0.004s time curl --unix-socket /go/src/github.com/docker/docker/bundles/test-integration-shell/docker.sock http://./containers/test/stats?stream=false\&one-shot=true > /dev/null 2>&1 real0m0.156s user0m0.007s sys0m0.007s ``` This fixes issues with downstreams ability to use the stats API to collect metrics. Signed-off-by: Brian Goff <cpuguy83@gmail.com>
2020-02-07 23:55:06 +00:00
if config.Stream && config.OneShot {
return errdefs.InvalidParameter(errors.New("cannot have stream=true and one-shot=true"))
}
// If the container is either not running or restarting and requires no stream, return an empty stats.
if (!ctr.IsRunning() || ctr.IsRestarting()) && !config.Stream {
return json.NewEncoder(config.OutStream).Encode(&types.StatsJSON{
Name: ctr.Name,
ID: ctr.ID,
})
}
// Get container stats directly if OneShot is set
if config.OneShot {
stats, err := daemon.GetContainerStats(ctr)
if err != nil {
return err
}
return json.NewEncoder(config.OutStream).Encode(stats)
}
outStream := config.OutStream
if config.Stream {
wf := ioutils.NewWriteFlusher(outStream)
defer wf.Close()
wf.Flush()
outStream = wf
}
var preCPUStats types.CPUStats
var preRead time.Time
getStatJSON := func(v interface{}) *types.StatsJSON {
ss := v.(types.StatsJSON)
ss.Name = ctr.Name
ss.ID = ctr.ID
ss.PreCPUStats = preCPUStats
ss.PreRead = preRead
preCPUStats = ss.CPUStats
preRead = ss.Read
return &ss
}
enc := json.NewEncoder(outStream)
updates := daemon.subscribeToContainerStats(ctr)
defer daemon.unsubscribeToContainerStats(ctr, updates)
Add stats options to not prime the stats Metrics collectors generally don't need the daemon to prime the stats with something to compare since they already have something to compare with. Before this change, the API does 2 collection cycles (which takes roughly 2s) in order to provide comparison for CPU usage over 1s. This was primarily added so that `docker stats --no-stream` had something to compare against. Really the CLI should have just made a 2nd call and done the comparison itself rather than forcing it on all API consumers. That ship has long sailed, though. With this change, clients can set an option to just pull a single stat, which is *at least* a full second faster: Old: ``` time curl --unix-socket /go/src/github.com/docker/docker/bundles/test-integration-shell/docker.sock http://./containers/test/stats?stream=false\&one-shot=false > /dev/null 2>&1 real0m1.864s user0m0.005s sys0m0.007s time curl --unix-socket /go/src/github.com/docker/docker/bundles/test-integration-shell/docker.sock http://./containers/test/stats?stream=false\&one-shot=false > /dev/null 2>&1 real0m1.173s user0m0.010s sys0m0.006s ``` New: ``` time curl --unix-socket /go/src/github.com/docker/docker/bundles/test-integration-shell/docker.sock http://./containers/test/stats?stream=false\&one-shot=true > /dev/null 2>&1 real0m0.680s user0m0.008s sys0m0.004s time curl --unix-socket /go/src/github.com/docker/docker/bundles/test-integration-shell/docker.sock http://./containers/test/stats?stream=false\&one-shot=true > /dev/null 2>&1 real0m0.156s user0m0.007s sys0m0.007s ``` This fixes issues with downstreams ability to use the stats API to collect metrics. Signed-off-by: Brian Goff <cpuguy83@gmail.com>
2020-02-07 23:55:06 +00:00
noStreamFirstFrame := !config.OneShot
for {
select {
case v, ok := <-updates:
if !ok {
return nil
}
statsJSON := getStatJSON(v)
if !config.Stream && noStreamFirstFrame {
// prime the cpu stats so they aren't 0 in the final output
noStreamFirstFrame = false
continue
}
if err := enc.Encode(statsJSON); err != nil {
return err
}
if !config.Stream {
return nil
}
case <-ctx.Done():
return nil
}
}
}
func (daemon *Daemon) subscribeToContainerStats(c *container.Container) chan interface{} {
return daemon.statsCollector.Collect(c)
}
func (daemon *Daemon) unsubscribeToContainerStats(c *container.Container, ch chan interface{}) {
daemon.statsCollector.Unsubscribe(c, ch)
}
// GetContainerStats collects all the stats published by a container
func (daemon *Daemon) GetContainerStats(container *container.Container) (*types.StatsJSON, error) {
stats, err := daemon.stats(container)
if err != nil {
goto done
}
// Sample system CPU usage close to container usage to avoid
// noise in metric calculations.
// FIXME: move to containerd on Linux (not Windows)
stats.CPUStats.SystemUsage, stats.CPUStats.OnlineCPUs, err = getSystemCPUUsage()
if err != nil {
goto done
}
// We already have the network stats on Windows directly from HCS.
if !container.Config.NetworkDisabled && runtime.GOOS != "windows" {
stats.Networks, err = daemon.getNetworkStats(container)
}
done:
switch err.(type) {
case nil:
return stats, nil
case errdefs.ErrConflict, errdefs.ErrNotFound:
// return empty stats containing only name and ID if not running or not found
return &types.StatsJSON{
Name: container.Name,
ID: container.ID,
}, nil
default:
log.G(context.TODO()).Errorf("collecting stats for container %s: %v", container.Name, err)
return nil, err
}
}