prometheus/storage/metric/rule_integration_test.go

954 lines
18 KiB
Go
Raw Normal View History

// Copyright 2013 Prometheus Team
// 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 metric
import (
2013-02-08 09:03:26 -08:00
"testing"
"time"
clientmodel "github.com/prometheus/client_golang/model"
"github.com/prometheus/prometheus/utility/test"
)
func GetValueAtTimeTests(persistenceMaker func() (ViewableMetricPersistence, test.Closer), t test.Tester) {
type value struct {
year int
month time.Month
day int
hour int
value clientmodel.SampleValue
}
type input struct {
year int
month time.Month
day int
hour int
}
type output []clientmodel.SampleValue
type behavior struct {
name string
input input
output output
}
var contexts = []struct {
name string
values []value
behaviors []behavior
}{
{
name: "no values",
values: []value{},
behaviors: []behavior{
{
name: "random target",
input: input{
year: 1984,
month: 3,
day: 30,
hour: 0,
},
},
},
},
{
name: "singleton",
values: []value{
{
year: 1984,
month: 3,
day: 30,
hour: 0,
value: 0,
},
},
behaviors: []behavior{
{
name: "exact",
input: input{
year: 1984,
month: 3,
day: 30,
hour: 0,
},
output: output{
0,
},
},
{
name: "before",
input: input{
year: 1984,
month: 3,
day: 29,
hour: 0,
},
output: output{
0,
},
},
{
name: "after",
input: input{
year: 1984,
month: 3,
day: 31,
hour: 0,
},
output: output{
0,
},
},
},
},
{
name: "double",
values: []value{
{
year: 1984,
month: 3,
day: 30,
hour: 0,
value: 0,
},
{
year: 1985,
month: 3,
day: 30,
hour: 0,
value: 1,
},
},
behaviors: []behavior{
{
name: "exact first",
input: input{
year: 1984,
month: 3,
day: 30,
hour: 0,
},
output: output{
0,
},
},
{
name: "exact second",
input: input{
year: 1985,
month: 3,
day: 30,
hour: 0,
},
output: output{
1,
},
},
{
name: "before first",
input: input{
year: 1983,
month: 9,
day: 29,
hour: 12,
},
output: output{
0,
},
},
{
name: "after second",
input: input{
year: 1985,
month: 9,
day: 28,
hour: 12,
},
output: output{
1,
},
},
{
name: "middle",
input: input{
year: 1984,
month: 9,
day: 28,
hour: 12,
},
output: output{
0,
1,
},
},
},
},
{
name: "triple",
values: []value{
{
year: 1984,
month: 3,
day: 30,
hour: 0,
value: 0,
},
{
year: 1985,
month: 3,
day: 30,
hour: 0,
value: 1,
},
{
year: 1986,
month: 3,
day: 30,
hour: 0,
value: 2,
},
},
behaviors: []behavior{
{
name: "exact first",
input: input{
year: 1984,
month: 3,
day: 30,
hour: 0,
},
output: output{
0,
},
},
{
name: "exact second",
input: input{
year: 1985,
month: 3,
day: 30,
hour: 0,
},
output: output{
1,
},
},
{
name: "exact third",
input: input{
year: 1986,
month: 3,
day: 30,
hour: 0,
},
output: output{
2,
},
},
{
name: "before first",
input: input{
year: 1983,
month: 9,
day: 29,
hour: 12,
},
output: output{
0,
},
},
{
name: "after third",
input: input{
year: 1986,
month: 9,
day: 28,
hour: 12,
},
output: output{
2,
},
},
{
name: "first middle",
input: input{
year: 1984,
month: 9,
day: 28,
hour: 12,
},
output: output{
0,
1,
},
},
{
name: "second middle",
input: input{
year: 1985,
month: 9,
day: 28,
hour: 12,
},
output: output{
1,
2,
},
},
},
},
}
for i, context := range contexts {
// Wrapping in function to enable garbage collection of resources.
func() {
p, closer := persistenceMaker()
defer closer.Close()
defer p.Close()
m := clientmodel.Metric{
clientmodel.MetricNameLabel: "age_in_years",
}
for _, value := range context.values {
testAppendSamples(p, &clientmodel.Sample{
Value: clientmodel.SampleValue(value.value),
Use custom timestamp type for sample timestamps and related code. So far we've been using Go's native time.Time for anything related to sample timestamps. Since the range of time.Time is much bigger than what we need, this has created two problems: - there could be time.Time values which were out of the range/precision of the time type that we persist to disk, therefore causing incorrectly ordered keys. One bug caused by this was: https://github.com/prometheus/prometheus/issues/367 It would be good to use a timestamp type that's more closely aligned with what the underlying storage supports. - sizeof(time.Time) is 192, while Prometheus should be ok with a single 64-bit Unix timestamp (possibly even a 32-bit one). Since we store samples in large numbers, this seriously affects memory usage. Furthermore, copying/working with the data will be faster if it's smaller. *MEMORY USAGE RESULTS* Initial memory usage comparisons for a running Prometheus with 1 timeseries and 100,000 samples show roughly a 13% decrease in total (VIRT) memory usage. In my tests, this advantage for some reason decreased a bit the more samples the timeseries had (to 5-7% for millions of samples). This I can't fully explain, but perhaps garbage collection issues were involved. *WHEN TO USE THE NEW TIMESTAMP TYPE* The new clientmodel.Timestamp type should be used whenever time calculations are either directly or indirectly related to sample timestamps. For example: - the timestamp of a sample itself - all kinds of watermarks - anything that may become or is compared to a sample timestamp (like the timestamp passed into Target.Scrape()). When to still use time.Time: - for measuring durations/times not related to sample timestamps, like duration telemetry exporting, timers that indicate how frequently to execute some action, etc. *NOTE ON OPERATOR OPTIMIZATION TESTS* We don't use operator optimization code anymore, but it still lives in the code as dead code. It still has tests, but I couldn't get all of them to pass with the new timestamp format. I commented out the failing cases for now, but we should probably remove the dead code soon. I just didn't want to do that in the same change as this. Change-Id: I821787414b0debe85c9fffaeb57abd453727af0f
2013-10-28 06:35:02 -07:00
Timestamp: clientmodel.TimestampFromTime(time.Date(value.year, value.month, value.day, value.hour, 0, 0, 0, time.UTC)),
Metric: m,
}, t)
}
for j, behavior := range context.behaviors {
input := behavior.input
Use custom timestamp type for sample timestamps and related code. So far we've been using Go's native time.Time for anything related to sample timestamps. Since the range of time.Time is much bigger than what we need, this has created two problems: - there could be time.Time values which were out of the range/precision of the time type that we persist to disk, therefore causing incorrectly ordered keys. One bug caused by this was: https://github.com/prometheus/prometheus/issues/367 It would be good to use a timestamp type that's more closely aligned with what the underlying storage supports. - sizeof(time.Time) is 192, while Prometheus should be ok with a single 64-bit Unix timestamp (possibly even a 32-bit one). Since we store samples in large numbers, this seriously affects memory usage. Furthermore, copying/working with the data will be faster if it's smaller. *MEMORY USAGE RESULTS* Initial memory usage comparisons for a running Prometheus with 1 timeseries and 100,000 samples show roughly a 13% decrease in total (VIRT) memory usage. In my tests, this advantage for some reason decreased a bit the more samples the timeseries had (to 5-7% for millions of samples). This I can't fully explain, but perhaps garbage collection issues were involved. *WHEN TO USE THE NEW TIMESTAMP TYPE* The new clientmodel.Timestamp type should be used whenever time calculations are either directly or indirectly related to sample timestamps. For example: - the timestamp of a sample itself - all kinds of watermarks - anything that may become or is compared to a sample timestamp (like the timestamp passed into Target.Scrape()). When to still use time.Time: - for measuring durations/times not related to sample timestamps, like duration telemetry exporting, timers that indicate how frequently to execute some action, etc. *NOTE ON OPERATOR OPTIMIZATION TESTS* We don't use operator optimization code anymore, but it still lives in the code as dead code. It still has tests, but I couldn't get all of them to pass with the new timestamp format. I commented out the failing cases for now, but we should probably remove the dead code soon. I just didn't want to do that in the same change as this. Change-Id: I821787414b0debe85c9fffaeb57abd453727af0f
2013-10-28 06:35:02 -07:00
time := clientmodel.TimestampFromTime(time.Date(input.year, input.month, input.day, input.hour, 0, 0, 0, time.UTC))
fingerprint := &clientmodel.Fingerprint{}
fingerprint.LoadFromMetric(m)
actual := p.GetValueAtTime(fingerprint, time)
if len(behavior.output) != len(actual) {
2013-05-21 09:12:02 -07:00
t.Fatalf("%d.%d(%s.%s). Expected %d samples but got: %v\n", i, j, context.name, behavior.name, len(behavior.output), actual)
}
for k, samplePair := range actual {
if samplePair.Value != behavior.output[k] {
2013-05-21 09:12:02 -07:00
t.Fatalf("%d.%d.%d(%s.%s). Expected %s but got %s\n", i, j, k, context.name, behavior.name, behavior.output[k], samplePair)
}
}
}
}()
}
}
func GetRangeValuesTests(persistenceMaker func() (ViewableMetricPersistence, test.Closer), onlyBoundaries bool, t test.Tester) {
type value struct {
year int
month time.Month
day int
hour int
value clientmodel.SampleValue
}
type input struct {
openYear int
openMonth time.Month
openDay int
openHour int
endYear int
endMonth time.Month
endDay int
endHour int
}
type output struct {
year int
month time.Month
day int
hour int
value clientmodel.SampleValue
}
type behavior struct {
name string
input input
output []output
}
var contexts = []struct {
name string
values []value
behaviors []behavior
}{
{
name: "no values",
values: []value{},
behaviors: []behavior{
{
name: "non-existent interval",
input: input{
openYear: 1984,
openMonth: 3,
openDay: 30,
openHour: 0,
endYear: 1985,
endMonth: 3,
endDay: 30,
endHour: 0,
},
},
},
},
{
name: "singleton value",
values: []value{
{
year: 1984,
month: 3,
day: 30,
hour: 0,
value: 0,
},
},
behaviors: []behavior{
{
name: "start on first value",
input: input{
openYear: 1984,
openMonth: 3,
openDay: 30,
openHour: 0,
endYear: 1985,
endMonth: 3,
endDay: 30,
endHour: 0,
},
output: []output{
{
year: 1984,
month: 3,
day: 30,
hour: 0,
value: 0,
},
},
},
{
name: "end on first value",
input: input{
openYear: 1983,
openMonth: 3,
openDay: 30,
openHour: 0,
endYear: 1984,
endMonth: 3,
endDay: 30,
endHour: 0,
},
output: []output{
{
year: 1984,
month: 3,
day: 30,
hour: 0,
value: 0,
},
},
},
{
name: "overlap on first value",
input: input{
openYear: 1983,
openMonth: 3,
openDay: 30,
openHour: 0,
endYear: 1985,
endMonth: 3,
endDay: 30,
endHour: 0,
},
output: []output{
{
year: 1984,
month: 3,
day: 30,
hour: 0,
value: 0,
},
},
},
},
},
{
name: "two values",
values: []value{
{
year: 1984,
month: 3,
day: 30,
hour: 0,
value: 0,
},
{
year: 1985,
month: 3,
day: 30,
hour: 0,
value: 1,
},
},
behaviors: []behavior{
{
name: "start on first value",
input: input{
openYear: 1984,
openMonth: 3,
openDay: 30,
openHour: 0,
endYear: 1985,
endMonth: 3,
endDay: 30,
endHour: 0,
},
output: []output{
{
year: 1984,
month: 3,
day: 30,
hour: 0,
value: 0,
},
{
year: 1985,
month: 3,
day: 30,
hour: 0,
value: 1,
},
},
},
{
name: "start on second value",
input: input{
openYear: 1985,
openMonth: 3,
openDay: 30,
openHour: 0,
endYear: 1986,
endMonth: 3,
endDay: 30,
endHour: 0,
},
output: []output{
{
year: 1985,
month: 3,
day: 30,
hour: 0,
value: 1,
},
},
},
{
name: "end on first value",
input: input{
openYear: 1983,
openMonth: 3,
openDay: 30,
openHour: 0,
endYear: 1984,
endMonth: 3,
endDay: 30,
endHour: 0,
},
output: []output{
{
year: 1984,
month: 3,
day: 30,
hour: 0,
value: 0,
},
},
},
{
name: "end on second value",
input: input{
openYear: 1985,
openMonth: 1,
openDay: 1,
openHour: 0,
endYear: 1985,
endMonth: 3,
endDay: 30,
endHour: 0,
},
output: []output{
{
year: 1985,
month: 3,
day: 30,
hour: 0,
value: 1,
},
},
},
{
name: "overlap on values",
input: input{
openYear: 1983,
openMonth: 3,
openDay: 30,
openHour: 0,
endYear: 1986,
endMonth: 3,
endDay: 30,
endHour: 0,
},
output: []output{
{
year: 1984,
month: 3,
day: 30,
hour: 0,
value: 0,
},
{
year: 1985,
month: 3,
day: 30,
hour: 0,
value: 1,
},
},
},
},
},
{
name: "three values",
values: []value{
{
year: 1984,
month: 3,
day: 30,
hour: 0,
value: 0,
},
{
year: 1985,
month: 3,
day: 30,
hour: 0,
value: 1,
},
{
year: 1986,
month: 3,
day: 30,
hour: 0,
value: 2,
},
},
behaviors: []behavior{
{
name: "start on first value",
input: input{
openYear: 1984,
openMonth: 3,
openDay: 30,
openHour: 0,
endYear: 1985,
endMonth: 3,
endDay: 30,
endHour: 0,
},
output: []output{
{
year: 1984,
month: 3,
day: 30,
hour: 0,
value: 0,
},
{
year: 1985,
month: 3,
day: 30,
hour: 0,
value: 1,
},
},
},
{
name: "start on second value",
input: input{
openYear: 1985,
openMonth: 3,
openDay: 30,
openHour: 0,
endYear: 1986,
endMonth: 3,
endDay: 30,
endHour: 0,
},
output: []output{
{
year: 1985,
month: 3,
day: 30,
hour: 0,
value: 1,
},
{
year: 1986,
month: 3,
day: 30,
hour: 0,
value: 2,
},
},
},
{
name: "end on first value",
input: input{
openYear: 1983,
openMonth: 3,
openDay: 30,
openHour: 0,
endYear: 1984,
endMonth: 3,
endDay: 30,
endHour: 0,
},
output: []output{
{
year: 1984,
month: 3,
day: 30,
hour: 0,
value: 0,
},
},
},
{
name: "end on second value",
input: input{
openYear: 1985,
openMonth: 1,
openDay: 1,
openHour: 0,
endYear: 1985,
endMonth: 3,
endDay: 30,
endHour: 0,
},
output: []output{
{
year: 1985,
month: 3,
day: 30,
hour: 0,
value: 1,
},
},
},
{
name: "overlap on values",
input: input{
openYear: 1983,
openMonth: 3,
openDay: 30,
openHour: 0,
endYear: 1986,
endMonth: 3,
endDay: 30,
endHour: 0,
},
output: []output{
{
year: 1984,
month: 3,
day: 30,
hour: 0,
value: 0,
},
{
year: 1985,
month: 3,
day: 30,
hour: 0,
value: 1,
},
{
year: 1986,
month: 3,
day: 30,
hour: 0,
value: 2,
},
},
},
},
},
}
for i, context := range contexts {
// Wrapping in function to enable garbage collection of resources.
func() {
p, closer := persistenceMaker()
defer closer.Close()
defer p.Close()
m := clientmodel.Metric{
clientmodel.MetricNameLabel: "age_in_years",
}
for _, value := range context.values {
testAppendSamples(p, &clientmodel.Sample{
Value: clientmodel.SampleValue(value.value),
Use custom timestamp type for sample timestamps and related code. So far we've been using Go's native time.Time for anything related to sample timestamps. Since the range of time.Time is much bigger than what we need, this has created two problems: - there could be time.Time values which were out of the range/precision of the time type that we persist to disk, therefore causing incorrectly ordered keys. One bug caused by this was: https://github.com/prometheus/prometheus/issues/367 It would be good to use a timestamp type that's more closely aligned with what the underlying storage supports. - sizeof(time.Time) is 192, while Prometheus should be ok with a single 64-bit Unix timestamp (possibly even a 32-bit one). Since we store samples in large numbers, this seriously affects memory usage. Furthermore, copying/working with the data will be faster if it's smaller. *MEMORY USAGE RESULTS* Initial memory usage comparisons for a running Prometheus with 1 timeseries and 100,000 samples show roughly a 13% decrease in total (VIRT) memory usage. In my tests, this advantage for some reason decreased a bit the more samples the timeseries had (to 5-7% for millions of samples). This I can't fully explain, but perhaps garbage collection issues were involved. *WHEN TO USE THE NEW TIMESTAMP TYPE* The new clientmodel.Timestamp type should be used whenever time calculations are either directly or indirectly related to sample timestamps. For example: - the timestamp of a sample itself - all kinds of watermarks - anything that may become or is compared to a sample timestamp (like the timestamp passed into Target.Scrape()). When to still use time.Time: - for measuring durations/times not related to sample timestamps, like duration telemetry exporting, timers that indicate how frequently to execute some action, etc. *NOTE ON OPERATOR OPTIMIZATION TESTS* We don't use operator optimization code anymore, but it still lives in the code as dead code. It still has tests, but I couldn't get all of them to pass with the new timestamp format. I commented out the failing cases for now, but we should probably remove the dead code soon. I just didn't want to do that in the same change as this. Change-Id: I821787414b0debe85c9fffaeb57abd453727af0f
2013-10-28 06:35:02 -07:00
Timestamp: clientmodel.TimestampFromTime(time.Date(value.year, value.month, value.day, value.hour, 0, 0, 0, time.UTC)),
Metric: m,
}, t)
}
for j, behavior := range context.behaviors {
input := behavior.input
Use custom timestamp type for sample timestamps and related code. So far we've been using Go's native time.Time for anything related to sample timestamps. Since the range of time.Time is much bigger than what we need, this has created two problems: - there could be time.Time values which were out of the range/precision of the time type that we persist to disk, therefore causing incorrectly ordered keys. One bug caused by this was: https://github.com/prometheus/prometheus/issues/367 It would be good to use a timestamp type that's more closely aligned with what the underlying storage supports. - sizeof(time.Time) is 192, while Prometheus should be ok with a single 64-bit Unix timestamp (possibly even a 32-bit one). Since we store samples in large numbers, this seriously affects memory usage. Furthermore, copying/working with the data will be faster if it's smaller. *MEMORY USAGE RESULTS* Initial memory usage comparisons for a running Prometheus with 1 timeseries and 100,000 samples show roughly a 13% decrease in total (VIRT) memory usage. In my tests, this advantage for some reason decreased a bit the more samples the timeseries had (to 5-7% for millions of samples). This I can't fully explain, but perhaps garbage collection issues were involved. *WHEN TO USE THE NEW TIMESTAMP TYPE* The new clientmodel.Timestamp type should be used whenever time calculations are either directly or indirectly related to sample timestamps. For example: - the timestamp of a sample itself - all kinds of watermarks - anything that may become or is compared to a sample timestamp (like the timestamp passed into Target.Scrape()). When to still use time.Time: - for measuring durations/times not related to sample timestamps, like duration telemetry exporting, timers that indicate how frequently to execute some action, etc. *NOTE ON OPERATOR OPTIMIZATION TESTS* We don't use operator optimization code anymore, but it still lives in the code as dead code. It still has tests, but I couldn't get all of them to pass with the new timestamp format. I commented out the failing cases for now, but we should probably remove the dead code soon. I just didn't want to do that in the same change as this. Change-Id: I821787414b0debe85c9fffaeb57abd453727af0f
2013-10-28 06:35:02 -07:00
open := clientmodel.TimestampFromTime(time.Date(input.openYear, input.openMonth, input.openDay, input.openHour, 0, 0, 0, time.UTC))
end := clientmodel.TimestampFromTime(time.Date(input.endYear, input.endMonth, input.endDay, input.endHour, 0, 0, 0, time.UTC))
in := Interval{
OldestInclusive: open,
NewestInclusive: end,
}
actualValues := Values{}
expectedValues := []output{}
fp := &clientmodel.Fingerprint{}
fp.LoadFromMetric(m)
if onlyBoundaries {
actualValues = p.GetBoundaryValues(fp, in)
l := len(behavior.output)
if l == 1 {
expectedValues = behavior.output[0:1]
}
if l > 1 {
expectedValues = append(behavior.output[0:1], behavior.output[l-1])
}
} else {
actualValues = p.GetRangeValues(fp, in)
expectedValues = behavior.output
}
if actualValues == nil && len(expectedValues) != 0 {
t.Fatalf("%d.%d(%s). Expected %v but got: %v\n", i, j, behavior.name, expectedValues, actualValues)
}
if expectedValues == nil {
if actualValues != nil {
t.Fatalf("%d.%d(%s). Expected nil values but got: %s\n", i, j, behavior.name, actualValues)
}
} else {
if len(expectedValues) != len(actualValues) {
t.Fatalf("%d.%d(%s). Expected length %d but got: %d\n", i, j, behavior.name, len(expectedValues), len(actualValues))
}
for k, actual := range actualValues {
expected := expectedValues[k]
if actual.Value != clientmodel.SampleValue(expected.value) {
t.Fatalf("%d.%d.%d(%s). Expected %v but got: %v\n", i, j, k, behavior.name, expected.value, actual.Value)
}
Use custom timestamp type for sample timestamps and related code. So far we've been using Go's native time.Time for anything related to sample timestamps. Since the range of time.Time is much bigger than what we need, this has created two problems: - there could be time.Time values which were out of the range/precision of the time type that we persist to disk, therefore causing incorrectly ordered keys. One bug caused by this was: https://github.com/prometheus/prometheus/issues/367 It would be good to use a timestamp type that's more closely aligned with what the underlying storage supports. - sizeof(time.Time) is 192, while Prometheus should be ok with a single 64-bit Unix timestamp (possibly even a 32-bit one). Since we store samples in large numbers, this seriously affects memory usage. Furthermore, copying/working with the data will be faster if it's smaller. *MEMORY USAGE RESULTS* Initial memory usage comparisons for a running Prometheus with 1 timeseries and 100,000 samples show roughly a 13% decrease in total (VIRT) memory usage. In my tests, this advantage for some reason decreased a bit the more samples the timeseries had (to 5-7% for millions of samples). This I can't fully explain, but perhaps garbage collection issues were involved. *WHEN TO USE THE NEW TIMESTAMP TYPE* The new clientmodel.Timestamp type should be used whenever time calculations are either directly or indirectly related to sample timestamps. For example: - the timestamp of a sample itself - all kinds of watermarks - anything that may become or is compared to a sample timestamp (like the timestamp passed into Target.Scrape()). When to still use time.Time: - for measuring durations/times not related to sample timestamps, like duration telemetry exporting, timers that indicate how frequently to execute some action, etc. *NOTE ON OPERATOR OPTIMIZATION TESTS* We don't use operator optimization code anymore, but it still lives in the code as dead code. It still has tests, but I couldn't get all of them to pass with the new timestamp format. I commented out the failing cases for now, but we should probably remove the dead code soon. I just didn't want to do that in the same change as this. Change-Id: I821787414b0debe85c9fffaeb57abd453727af0f
2013-10-28 06:35:02 -07:00
if actual.Timestamp.Time().Year() != expected.year {
t.Fatalf("%d.%d.%d(%s). Expected %d but got: %d\n", i, j, k, behavior.name, expected.year, actual.Timestamp.Time().Year())
}
Use custom timestamp type for sample timestamps and related code. So far we've been using Go's native time.Time for anything related to sample timestamps. Since the range of time.Time is much bigger than what we need, this has created two problems: - there could be time.Time values which were out of the range/precision of the time type that we persist to disk, therefore causing incorrectly ordered keys. One bug caused by this was: https://github.com/prometheus/prometheus/issues/367 It would be good to use a timestamp type that's more closely aligned with what the underlying storage supports. - sizeof(time.Time) is 192, while Prometheus should be ok with a single 64-bit Unix timestamp (possibly even a 32-bit one). Since we store samples in large numbers, this seriously affects memory usage. Furthermore, copying/working with the data will be faster if it's smaller. *MEMORY USAGE RESULTS* Initial memory usage comparisons for a running Prometheus with 1 timeseries and 100,000 samples show roughly a 13% decrease in total (VIRT) memory usage. In my tests, this advantage for some reason decreased a bit the more samples the timeseries had (to 5-7% for millions of samples). This I can't fully explain, but perhaps garbage collection issues were involved. *WHEN TO USE THE NEW TIMESTAMP TYPE* The new clientmodel.Timestamp type should be used whenever time calculations are either directly or indirectly related to sample timestamps. For example: - the timestamp of a sample itself - all kinds of watermarks - anything that may become or is compared to a sample timestamp (like the timestamp passed into Target.Scrape()). When to still use time.Time: - for measuring durations/times not related to sample timestamps, like duration telemetry exporting, timers that indicate how frequently to execute some action, etc. *NOTE ON OPERATOR OPTIMIZATION TESTS* We don't use operator optimization code anymore, but it still lives in the code as dead code. It still has tests, but I couldn't get all of them to pass with the new timestamp format. I commented out the failing cases for now, but we should probably remove the dead code soon. I just didn't want to do that in the same change as this. Change-Id: I821787414b0debe85c9fffaeb57abd453727af0f
2013-10-28 06:35:02 -07:00
if actual.Timestamp.Time().Month() != expected.month {
t.Fatalf("%d.%d.%d(%s). Expected %d but got: %d\n", i, j, k, behavior.name, expected.month, actual.Timestamp.Time().Month())
}
// XXX: Find problem here.
// Mismatches occur in this and have for a long time in the LevelDB
// case, however not im-memory.
//
// if actual.Timestamp.Day() != expected.day {
// t.Fatalf("%d.%d.%d(%s). Expected %d but got: %d\n", i, j, k, behavior.name, expected.day, actual.Timestamp.Day())
// }
// if actual.Timestamp.Hour() != expected.hour {
// t.Fatalf("%d.%d.%d(%s). Expected %d but got: %d\n", i, j, k, behavior.name, expected.hour, actual.Timestamp.Hour())
// }
}
}
}
}()
}
}
2013-02-08 09:03:26 -08:00
// Test Definitions Follow
func testMemoryGetValueAtTime(t test.Tester) {
persistenceMaker := func() (ViewableMetricPersistence, test.Closer) {
return NewMemorySeriesStorage(MemorySeriesOptions{}), test.NilCloser
2013-02-08 09:03:26 -08:00
}
GetValueAtTimeTests(persistenceMaker, t)
}
func TestMemoryGetValueAtTime(t *testing.T) {
testMemoryGetValueAtTime(t)
}
func BenchmarkMemoryGetValueAtTime(b *testing.B) {
for i := 0; i < b.N; i++ {
testMemoryGetValueAtTime(b)
}
}
func TestMemoryGetBoundaryValues(t *testing.T) {
testMemoryGetBoundaryValues(t)
}
func BenchmarkMemoryGetBoundaryValues(b *testing.B) {
for i := 0; i < b.N; i++ {
testMemoryGetBoundaryValues(b)
}
}
func testMemoryGetRangeValues(t test.Tester) {
persistenceMaker := func() (ViewableMetricPersistence, test.Closer) {
return NewMemorySeriesStorage(MemorySeriesOptions{}), test.NilCloser
2013-02-08 09:03:26 -08:00
}
GetRangeValuesTests(persistenceMaker, false, t)
}
func testMemoryGetBoundaryValues(t test.Tester) {
persistenceMaker := func() (ViewableMetricPersistence, test.Closer) {
return NewMemorySeriesStorage(MemorySeriesOptions{}), test.NilCloser
}
GetRangeValuesTests(persistenceMaker, true, t)
2013-02-08 09:03:26 -08:00
}
func TestMemoryGetRangeValues(t *testing.T) {
testMemoryGetRangeValues(t)
}
func BenchmarkMemoryGetRangeValues(b *testing.B) {
for i := 0; i < b.N; i++ {
testMemoryGetRangeValues(b)
}
}