README
¶
go-metrics
Go port of Coda Hale's Metrics library: https://github.com/codahale/metrics.
This code is not safe on 32-bit architectures. It will be as soon as atomic.LoadInt64 lands in a release tag.
Usage
Create and update metrics:
r := metrics.NewRegistry()
c := metrics.NewCounter()
r.RegisterCounter("foo", c)
c.Inc(47)
g := metrics.NewGauge()
r.RegisterGauge("bar", g)
g.Update(47)
s := metrics.NewExpDecaySample(1028, 0.015)
//s := metrics.NewUniformSample(1028)
h := metrics.NewHistogram(s)
r.RegisterHistogram("baz", h)
h.Update(47)
m := metrics.NewMeter()
r.RegisterMeter("quux", m)
m.Mark(47)
t := metrics.NewTimer()
r.RegisterTimer("bang", t)
t.Update(47)
t.Time(func() {})
Periodically log every metric in human-readable form to standard error:
metrics.Log(r, 60, log.New(os.Stderr, "metrics: ", log.Lmicroseconds))
Periodically log every metric in slightly-more-parseable form to syslog:
w, err := syslog.Dial("unixgram", "/dev/log", syslog.LOG_INFO, "metrics")
if nil != err { log.Fatalln(err) }
metrics.Syslog(r, 60, w)
Installation
goinstall github.com/rcrowley/go-metrics
Documentation
¶
Overview ¶
Go port of Coda Hale's Metrics library
<https://github.com/rcrowley/go-metrics>
Coda Hale's original work: <https://github.com/codahale/metrics>
Index ¶
- func CaptureRuntimeMemStats(r Registry, updateMemStats bool)
- func Log(r Registry, interval int, l *log.Logger)
- func RegisterRuntimeMemStats(r Registry)
- func Syslog(r Registry, interval int, w *syslog.Writer)
- type Counter
- type EWMA
- type ExpDecaySample
- type Gauge
- type Healthcheck
- type Histogram
- type Meter
- type Registry
- type Sample
- type StandardCounter
- type StandardEWMA
- type StandardGauge
- type StandardHealthcheck
- type StandardHistogram
- func (h *StandardHistogram) Clear()
- func (h *StandardHistogram) Count() int64
- func (h *StandardHistogram) Max() int64
- func (h *StandardHistogram) Mean() float64
- func (h *StandardHistogram) Min() int64
- func (h *StandardHistogram) Percentile(p float64) float64
- func (h *StandardHistogram) Percentiles(ps []float64) []float64
- func (h *StandardHistogram) StdDev() float64
- func (h *StandardHistogram) Update(v int64)
- func (h *StandardHistogram) Variance() float64
- type StandardMeter
- type StandardRegistry
- type StandardTimer
- func (t *StandardTimer) Count() int64
- func (t *StandardTimer) Max() int64
- func (t *StandardTimer) Mean() float64
- func (t *StandardTimer) Min() int64
- func (t *StandardTimer) Percentile(p float64) float64
- func (t *StandardTimer) Percentiles(ps []float64) []float64
- func (t *StandardTimer) Rate1() float64
- func (t *StandardTimer) Rate15() float64
- func (t *StandardTimer) Rate5() float64
- func (t *StandardTimer) RateMean() float64
- func (t *StandardTimer) StdDev() float64
- func (t *StandardTimer) Time(f func())
- func (t *StandardTimer) Update(duration uint64)
- type Timer
- type UniformSample
Constants ¶
This section is empty.
Variables ¶
This section is empty.
Functions ¶
func CaptureRuntimeMemStats ¶
Capture new values for the Go runtime statistics exported in runtime.MemStats. This is designed to be called in a background goroutine. Giving a registry which has not been given to RegisterRuntimeMemStats will panic. If the second parameter is false, the counters will be left to the lazy updates provided by the runtime.
func Log ¶
Output each metric in the given registry periodically using the given logger. The interval is to be given in seconds.
func RegisterRuntimeMemStats ¶
func RegisterRuntimeMemStats(r Registry)
Register metrics for the Go runtime statistics exported in runtime.MemStats. The metrics are named by their fully-qualified Go symbols, i.e. runtime.MemStatsAlloc. In addition to runtime.MemStats, register the return value of runtime.Goroutines() as runtime.Goroutines.
Types ¶
type Counter ¶
Counters hold an int64 value that can be incremented and decremented.
This is an interface so as to encourage other structs to implement the Counter API as appropriate.
type EWMA ¶
EWMAs continuously calculate an exponentially-weighted moving average based on an outside source of clock ticks.
This is an interface so as to encourage other structs to implement the EWMA API as appropriate.
type ExpDecaySample ¶
type ExpDecaySample struct {
// contains filtered or unexported fields
}
An exponentially-decaying sample using a forward-decaying priority reservoir. See Cormode et al's "Forward Decay: A Practical Time Decay Model for Streaming Systems".
func NewExpDecaySample ¶
func NewExpDecaySample(reservoirSize int, alpha float64) *ExpDecaySample
Create a new exponentially-decaying sample with the given reservoir size and alpha.
func (*ExpDecaySample) Size ¶
func (s *ExpDecaySample) Size() int
Return the size of the sample, which is at most the reservoir size.
func (*ExpDecaySample) Update ¶
func (s *ExpDecaySample) Update(v int64)
Update the sample with a new value.
func (*ExpDecaySample) Values ¶
func (s *ExpDecaySample) Values() []int64
Return all the values in the sample.
type Gauge ¶
Gauges hold an int64 value that can be set arbitrarily.
This is an interface so as to encourage other structs to implement the Gauge API as appropriate.
type Healthcheck ¶
Healthchecks hold an os.Error value describing an arbitrary up/down status.
This is an interface so as to encourage other structs to implement the Healthcheck API as appropriate.
type Histogram ¶
type Histogram interface {
Clear()
Count() int64
Max() int64
Mean() float64
Min() int64
Percentile(float64) float64
Percentiles([]float64) []float64
StdDev() float64
Update(int64)
Variance() float64
}
Histograms calculate distribution statistics from an int64 value.
This is an interface so as to encourage other structs to implement the Histogram API as appropriate.
type Meter ¶
type Meter interface {
Count() int64
Mark(int64)
Rate1() float64
Rate5() float64
Rate15() float64
RateMean() float64
}
Meters count events to produce exponentially-weighted moving average rates at one-, five-, and fifteen-minutes and a mean rate.
This is an interface so as to encourage other structs to implement the Meter API as appropriate.
type Registry ¶
type Registry interface {
Each(func(string, interface{}))
Get(string) interface{}
Register(string, interface{})
RunHealthchecks()
Unregister(string)
}
A Registry holds references to a set of metrics by name and can iterate over them, calling callback functions provided by the user.
This is an interface so as to encourage other structs to implement the Registry API as appropriate.
type Sample ¶
Samples maintain a statistically-significant selection of values from a stream.
This is an interface so as to encourage other structs to implement the Sample API as appropriate.
type StandardCounter ¶
type StandardCounter struct {
// contains filtered or unexported fields
}
The standard implementation of a Counter uses the sync/atomic package to manage a single int64 value. When the latest weeklies land in a release, atomic.LoadInt64 will be available and this code will become safe on 32-bit architectures.
func (*StandardCounter) Clear ¶
func (c *StandardCounter) Clear()
Clear the counter: set it to zero.
func (*StandardCounter) Count ¶
func (c *StandardCounter) Count() int64
Return the current count. This is the method that's currently unsafe on 32-bit architectures.
func (*StandardCounter) Dec ¶
func (c *StandardCounter) Dec(i int64)
Decrement the counter by the given amount.
func (*StandardCounter) Inc ¶
func (c *StandardCounter) Inc(i int64)
Increment the counter by the given amount.
type StandardEWMA ¶
type StandardEWMA struct {
// contains filtered or unexported fields
}
The standard implementation of an EWMA tracks the number of uncounted events and processes them on each tick. It uses the sync/atomic package to manage uncounted events. When the latest weeklies land in a release, atomic.LoadInt64 will be available and this code will become safe on 32-bit architectures.
func NewEWMA ¶
func NewEWMA(alpha float64) *StandardEWMA
Create a new EWMA with the given alpha. Create the clock channel and start the ticker goroutine.
func NewEWMA1 ¶
func NewEWMA1() *StandardEWMA
Create a new EWMA with alpha set for a one-minute moving average.
func NewEWMA15 ¶
func NewEWMA15() *StandardEWMA
Create a new EWMA with alpha set for a fifteen-minute moving average.
func NewEWMA5 ¶
func NewEWMA5() *StandardEWMA
Create a new EWMA with alpha set for a five-minute moving average.
func (*StandardEWMA) Rate ¶
func (a *StandardEWMA) Rate() float64
Return the moving average rate of events per second.
func (*StandardEWMA) Tick ¶
func (a *StandardEWMA) Tick()
Tick the clock to update the moving average.
type StandardGauge ¶
type StandardGauge struct {
// contains filtered or unexported fields
}
The standard implementation of a Gauge uses the sync/atomic package to manage a single int64 value. When the latest weeklies land in a release, atomic.LoadInt64 will be available and this code will become safe on 32-bit architectures.
func (*StandardGauge) Value ¶
func (g *StandardGauge) Value() int64
Return the gauge's current value.
type StandardHealthcheck ¶
type StandardHealthcheck struct {
// contains filtered or unexported fields
}
The standard implementation of a Healthcheck stores the status and a function to call to update the status.
func NewHealthcheck ¶
func NewHealthcheck(f func(Healthcheck)) *StandardHealthcheck
Create a new healthcheck, which will use the given function to update its status.
func (*StandardHealthcheck) Check ¶
func (h *StandardHealthcheck) Check()
Update the healthcheck's status.
func (*StandardHealthcheck) Error ¶
func (h *StandardHealthcheck) Error() error
Return the healthcheck's status, which will be nil if it is healthy.
func (*StandardHealthcheck) Healthy ¶
func (h *StandardHealthcheck) Healthy()
Mark the healthcheck as healthy.
func (*StandardHealthcheck) Unhealthy ¶
func (h *StandardHealthcheck) Unhealthy(err error)
Mark the healthcheck as unhealthy. The error should provide details.
type StandardHistogram ¶
type StandardHistogram struct {
// contains filtered or unexported fields
}
The standard implementation of a Histogram uses a Sample and a goroutine to synchronize its calculations.
func NewHistogram ¶
func NewHistogram(s Sample) *StandardHistogram
Create a new histogram with the given Sample. Create the communication channels and start the synchronizing goroutine.
func (*StandardHistogram) Count ¶
func (h *StandardHistogram) Count() int64
Return the count of inputs since the histogram was last cleared.
func (*StandardHistogram) Max ¶
func (h *StandardHistogram) Max() int64
Return the maximal value seen since the histogram was last cleared.
func (*StandardHistogram) Mean ¶
func (h *StandardHistogram) Mean() float64
Return the mean of all values seen since the histogram was last cleared.
func (*StandardHistogram) Min ¶
func (h *StandardHistogram) Min() int64
Return the minimal value seen since the histogram was last cleared.
func (*StandardHistogram) Percentile ¶
func (h *StandardHistogram) Percentile(p float64) float64
Return an arbitrary percentile of all values seen since the histogram was last cleared.
func (*StandardHistogram) Percentiles ¶
func (h *StandardHistogram) Percentiles(ps []float64) []float64
Return a slice of arbitrary percentiles of all values seen since the histogram was last cleared.
func (*StandardHistogram) StdDev ¶
func (h *StandardHistogram) StdDev() float64
Return the standard deviation of all values seen since the histogram was last cleared.
func (*StandardHistogram) Update ¶
func (h *StandardHistogram) Update(v int64)
Update the histogram with a new value.
func (*StandardHistogram) Variance ¶
func (h *StandardHistogram) Variance() float64
Return the variance of all values seen since the histogram was last cleared.
type StandardMeter ¶
type StandardMeter struct {
// contains filtered or unexported fields
}
The standard implementation of a Meter uses a goroutine to synchronize its calculations and another goroutine (via time.Ticker) to produce clock ticks.
func NewMeter ¶
func NewMeter() *StandardMeter
Create a new meter. Create the communication channels and start the synchronizing goroutine.
func (*StandardMeter) Count ¶
func (m *StandardMeter) Count() int64
Return the count of events seen.
func (*StandardMeter) Rate1 ¶
func (m *StandardMeter) Rate1() float64
Return the meter's one-minute moving average rate of events.
func (*StandardMeter) Rate15 ¶
func (m *StandardMeter) Rate15() float64
Return the meter's fifteen-minute moving average rate of events.
func (*StandardMeter) Rate5 ¶
func (m *StandardMeter) Rate5() float64
Return the meter's five-minute moving average rate of events.
func (*StandardMeter) RateMean ¶
func (m *StandardMeter) RateMean() float64
Return the meter's mean rate of events.
type StandardRegistry ¶
type StandardRegistry struct {
// contains filtered or unexported fields
}
The standard implementation of a Registry is a set of mutex-protected maps of names to metrics.
func (*StandardRegistry) Each ¶
func (r *StandardRegistry) Each(f func(string, interface{}))
Call the given function for each registered metric.
func (*StandardRegistry) Get ¶
func (r *StandardRegistry) Get(name string) interface{}
Get the metric by the given name or nil if none is registered.
func (*StandardRegistry) Register ¶
func (r *StandardRegistry) Register(name string, i interface{})
Register the given metric under the given name.
func (*StandardRegistry) RunHealthchecks ¶
func (r *StandardRegistry) RunHealthchecks()
Run all registered healthchecks.
func (*StandardRegistry) Unregister ¶
func (r *StandardRegistry) Unregister(name string)
Unregister the metric with the given name.
type StandardTimer ¶
type StandardTimer struct {
// contains filtered or unexported fields
}
The standard implementation of a Timer uses a Histogram and Meter directly.
func NewCustomTimer ¶
func NewCustomTimer(h Histogram, m Meter) *StandardTimer
Create a new timer with the given Histogram and Meter.
func NewTimer ¶
func NewTimer() *StandardTimer
Create a new timer with a standard histogram and meter. The histogram will use an exponentially-decaying sample with the same reservoir size and alpha as UNIX load averages.
func (*StandardTimer) Mean ¶
func (t *StandardTimer) Mean() float64
Return the mean of all values seen.
func (*StandardTimer) Percentile ¶
func (t *StandardTimer) Percentile(p float64) float64
Return an arbitrary percentile of all values seen.
func (*StandardTimer) Percentiles ¶
func (t *StandardTimer) Percentiles(ps []float64) []float64
Return a slice of arbitrary percentiles of all values seen.
func (*StandardTimer) Rate1 ¶
func (t *StandardTimer) Rate1() float64
Return the meter's one-minute moving average rate of events.
func (*StandardTimer) Rate15 ¶
func (t *StandardTimer) Rate15() float64
Return the meter's fifteen-minute moving average rate of events.
func (*StandardTimer) Rate5 ¶
func (t *StandardTimer) Rate5() float64
Return the meter's five-minute moving average rate of events.
func (*StandardTimer) RateMean ¶
func (t *StandardTimer) RateMean() float64
Return the meter's mean rate of events.
func (*StandardTimer) StdDev ¶
func (t *StandardTimer) StdDev() float64
Return the standard deviation of all values seen.
func (*StandardTimer) Time ¶
func (t *StandardTimer) Time(f func())
Record the duration of the execution of the given function.
func (*StandardTimer) Update ¶
func (t *StandardTimer) Update(duration uint64)
Record the duration of an event.
type Timer ¶
type Timer interface {
Count() int64
Max() int64
Mean() float64
Min() int64
Percentile(float64) float64
Percentiles([]float64) []float64
Rate1() float64
Rate5() float64
Rate15() float64
RateMean() float64
StdDev() float64
Time(func())
Update(uint64)
}
Timers capture the duration and rate of events.
This is an interface so as to encourage other structs to implement the Histogram API as appropriate.
type UniformSample ¶
type UniformSample struct {
// contains filtered or unexported fields
}
A uniform sample using Vitter's Algorithm R.
<http://www.cs.umd.edu/~samir/498/vitter.pdf>
func NewUniformSample ¶
func NewUniformSample(reservoirSize int) *UniformSample
Create a new uniform sample with the given reservoir size.
func (*UniformSample) Size ¶
func (s *UniformSample) Size() int
Return the size of the sample, which is at most the reservoir size.
func (*UniformSample) Update ¶
func (s *UniformSample) Update(v int64)
Update the sample with a new value.
func (*UniformSample) Values ¶
func (s *UniformSample) Values() []int64
Return all the values in the sample.
Source Files
Directories