client_golang/prometheus/go_collector_go116.go

123 lines
3.6 KiB
Go
Raw Permalink Normal View History

Use the runtime/metrics package for the Go collector for 1.17+ (#955) This change introduces use of the runtime/metrics package in place of runtime.MemStats for Go 1.17 or later. The runtime/metrics package was introduced in Go 1.16, but not all the old metrics were accounted for until 1.17. The runtime/metrics package offers several advantages over using runtime.MemStats: * The list of metrics and their descriptions are machine-readable, allowing new metrics to get added without any additional work. * Detailed histogram-based metrics are now available, offering much deeper insights into the Go runtime. * The runtime/metrics API is significantly more efficient than runtime.MemStats, even with the additional metrics added, because it does not require any stop-the-world events. That being said, integrating the package comes with some caveats, some of which were discussed in #842. Namely: * The old MemStats-based metrics need to continue working, so they're exported under their old names backed by equivalent runtime/metrics metrics. * Earlier versions of Go need to continue working, so the old code remains, but behind a build tag. Finally, a few notes about the implementation: * This change includes a whole bunch of refactoring to avoid significant code duplication. * This change adds a new histogram metric type specifically optimized for runtime/metrics histograms. This type's methods also include additional logic to deal with differences in bounds conventions. * This change makes a whole bunch of decisions about how runtime/metrics names are translated. * This change adds a `go generate` script to generate a list of expected runtime/metrics names for a given Go version for auditing. Users of new versions of Go will transparently be allowed to use new metrics, however. Signed-off-by: Michael Anthony Knyszek <mknyszek@google.com>
2022-01-16 19:41:56 +03:00
// Copyright 2021 The Prometheus 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.
//go:build !go1.17
// +build !go1.17
package prometheus
import (
"runtime"
"sync"
"time"
)
type goCollector struct {
base baseGoCollector
// ms... are memstats related.
msLast *runtime.MemStats // Previously collected memstats.
msLastTimestamp time.Time
msMtx sync.Mutex // Protects msLast and msLastTimestamp.
msMetrics memStatsMetrics
msRead func(*runtime.MemStats) // For mocking in tests.
msMaxWait time.Duration // Wait time for fresh memstats.
msMaxAge time.Duration // Maximum allowed age of old memstats.
}
// NewGoCollector is the obsolete version of collectors.NewGoCollector.
// See there for documentation.
//
// Deprecated: Use collectors.NewGoCollector instead.
func NewGoCollector() Collector {
msMetrics := goRuntimeMemStats()
msMetrics = append(msMetrics, struct {
desc *Desc
eval func(*runtime.MemStats) float64
valType ValueType
}{
// This metric is omitted in Go1.17+, see https://github.com/prometheus/client_golang/issues/842#issuecomment-861812034
desc: NewDesc(
memstatNamespace("gc_cpu_fraction"),
"The fraction of this program's available CPU time used by the GC since the program started.",
nil, nil,
),
eval: func(ms *runtime.MemStats) float64 { return ms.GCCPUFraction },
valType: GaugeValue,
})
Use the runtime/metrics package for the Go collector for 1.17+ (#955) This change introduces use of the runtime/metrics package in place of runtime.MemStats for Go 1.17 or later. The runtime/metrics package was introduced in Go 1.16, but not all the old metrics were accounted for until 1.17. The runtime/metrics package offers several advantages over using runtime.MemStats: * The list of metrics and their descriptions are machine-readable, allowing new metrics to get added without any additional work. * Detailed histogram-based metrics are now available, offering much deeper insights into the Go runtime. * The runtime/metrics API is significantly more efficient than runtime.MemStats, even with the additional metrics added, because it does not require any stop-the-world events. That being said, integrating the package comes with some caveats, some of which were discussed in #842. Namely: * The old MemStats-based metrics need to continue working, so they're exported under their old names backed by equivalent runtime/metrics metrics. * Earlier versions of Go need to continue working, so the old code remains, but behind a build tag. Finally, a few notes about the implementation: * This change includes a whole bunch of refactoring to avoid significant code duplication. * This change adds a new histogram metric type specifically optimized for runtime/metrics histograms. This type's methods also include additional logic to deal with differences in bounds conventions. * This change makes a whole bunch of decisions about how runtime/metrics names are translated. * This change adds a `go generate` script to generate a list of expected runtime/metrics names for a given Go version for auditing. Users of new versions of Go will transparently be allowed to use new metrics, however. Signed-off-by: Michael Anthony Knyszek <mknyszek@google.com>
2022-01-16 19:41:56 +03:00
return &goCollector{
base: newBaseGoCollector(),
msLast: &runtime.MemStats{},
msRead: runtime.ReadMemStats,
msMaxWait: time.Second,
msMaxAge: 5 * time.Minute,
msMetrics: msMetrics,
Use the runtime/metrics package for the Go collector for 1.17+ (#955) This change introduces use of the runtime/metrics package in place of runtime.MemStats for Go 1.17 or later. The runtime/metrics package was introduced in Go 1.16, but not all the old metrics were accounted for until 1.17. The runtime/metrics package offers several advantages over using runtime.MemStats: * The list of metrics and their descriptions are machine-readable, allowing new metrics to get added without any additional work. * Detailed histogram-based metrics are now available, offering much deeper insights into the Go runtime. * The runtime/metrics API is significantly more efficient than runtime.MemStats, even with the additional metrics added, because it does not require any stop-the-world events. That being said, integrating the package comes with some caveats, some of which were discussed in #842. Namely: * The old MemStats-based metrics need to continue working, so they're exported under their old names backed by equivalent runtime/metrics metrics. * Earlier versions of Go need to continue working, so the old code remains, but behind a build tag. Finally, a few notes about the implementation: * This change includes a whole bunch of refactoring to avoid significant code duplication. * This change adds a new histogram metric type specifically optimized for runtime/metrics histograms. This type's methods also include additional logic to deal with differences in bounds conventions. * This change makes a whole bunch of decisions about how runtime/metrics names are translated. * This change adds a `go generate` script to generate a list of expected runtime/metrics names for a given Go version for auditing. Users of new versions of Go will transparently be allowed to use new metrics, however. Signed-off-by: Michael Anthony Knyszek <mknyszek@google.com>
2022-01-16 19:41:56 +03:00
}
}
// Describe returns all descriptions of the collector.
func (c *goCollector) Describe(ch chan<- *Desc) {
c.base.Describe(ch)
for _, i := range c.msMetrics {
ch <- i.desc
}
}
// Collect returns the current state of all metrics of the collector.
func (c *goCollector) Collect(ch chan<- Metric) {
var (
ms = &runtime.MemStats{}
done = make(chan struct{})
)
// Start reading memstats first as it might take a while.
go func() {
c.msRead(ms)
c.msMtx.Lock()
c.msLast = ms
c.msLastTimestamp = time.Now()
c.msMtx.Unlock()
close(done)
}()
// Collect base non-memory metrics.
c.base.Collect(ch)
timer := time.NewTimer(c.msMaxWait)
select {
case <-done: // Our own ReadMemStats succeeded in time. Use it.
timer.Stop() // Important for high collection frequencies to not pile up timers.
c.msCollect(ch, ms)
return
case <-timer.C: // Time out, use last memstats if possible. Continue below.
}
c.msMtx.Lock()
if time.Since(c.msLastTimestamp) < c.msMaxAge {
// Last memstats are recent enough. Collect from them under the lock.
c.msCollect(ch, c.msLast)
c.msMtx.Unlock()
return
}
// If we are here, the last memstats are too old or don't exist. We have
// to wait until our own ReadMemStats finally completes. For that to
// happen, we have to release the lock.
c.msMtx.Unlock()
<-done
c.msCollect(ch, ms)
}
func (c *goCollector) msCollect(ch chan<- Metric, ms *runtime.MemStats) {
for _, i := range c.msMetrics {
ch <- MustNewConstMetric(i.desc, i.valType, i.eval(ms))
}
}