The Prometheus monitoring system and time series database.
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

653 lines
19 KiB

// Copyright 2014 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.
package template
import (
"context"
"math"
"net/url"
"reflect"
"testing"
"time"
"github.com/stretchr/testify/require"
"github.com/prometheus/prometheus/model/histogram"
"github.com/prometheus/prometheus/model/labels"
"github.com/prometheus/prometheus/promql"
)
func TestTemplateExpansion(t *testing.T) {
testTemplateExpansion(t, []scenario{
{
// No template.
text: "plain text",
output: "plain text",
},
{
// Simple value.
text: "{{ 1 }}",
output: "1",
},
{
// Native histogram value.
text: "{{ . | value }}",
input: &sample{Value: &histogram.FloatHistogram{Count: 3, Sum: 10}},
output: (&histogram.FloatHistogram{Count: 3, Sum: 10}).String(),
},
{
// Non-ASCII space (not allowed in text/template, see https://github.com/golang/go/blob/master/src/text/template/parse/lex.go#L98)
text: "{{ }}",
shouldFail: true,
errorMsg: "error parsing template test: template: test:1: unrecognized character in action: U+00A0",
},
{
// HTML escaping.
text: "{{ \"<b>\" }}",
output: "&lt;b&gt;",
html: true,
},
{
// Disabling HTML escaping.
text: "{{ \"<b>\" | safeHtml }}",
output: "<b>",
html: true,
},
{
// HTML escaping doesn't apply to non-html.
text: "{{ \"<b>\" }}",
output: "<b>",
},
{
// Pass multiple arguments to templates.
text: "{{define \"x\"}}{{.arg0}} {{.arg1}}{{end}}{{template \"x\" (args 1 \"2\")}}",
output: "1 2",
},
{
text: "{{ query \"1.5\" | first | value }}",
output: "1.5",
promql: Separate `Point` into `FPoint` and `HPoint` In other words: Instead of having a “polymorphous” `Point` that can either contain a float value or a histogram value, use an `FPoint` for floats and an `HPoint` for histograms. This seemingly small change has a _lot_ of repercussions throughout the codebase. The idea here is to avoid the increase in size of `Point` arrays that happened after native histograms had been added. The higher-level data structures (`Sample`, `Series`, etc.) are still “polymorphous”. The same idea could be applied to them, but at each step the trade-offs needed to be evaluated. The idea with this change is to do the minimum necessary to get back to pre-histogram performance for functions that do not touch histograms. Here are comparisons for the `changes` function. The test data doesn't include histograms yet. Ideally, there would be no change in the benchmark result at all. First runtime v2.39 compared to directly prior to this commit: ``` name old time/op new time/op delta RangeQuery/expr=changes(a_one[1d]),steps=1-16 391µs ± 2% 542µs ± 1% +38.58% (p=0.000 n=9+8) RangeQuery/expr=changes(a_one[1d]),steps=10-16 452µs ± 2% 617µs ± 2% +36.48% (p=0.000 n=10+10) RangeQuery/expr=changes(a_one[1d]),steps=100-16 1.12ms ± 1% 1.36ms ± 2% +21.58% (p=0.000 n=8+10) RangeQuery/expr=changes(a_one[1d]),steps=1000-16 7.83ms ± 1% 8.94ms ± 1% +14.21% (p=0.000 n=10+10) RangeQuery/expr=changes(a_ten[1d]),steps=1-16 2.98ms ± 0% 3.30ms ± 1% +10.67% (p=0.000 n=9+10) RangeQuery/expr=changes(a_ten[1d]),steps=10-16 3.66ms ± 1% 4.10ms ± 1% +11.82% (p=0.000 n=10+10) RangeQuery/expr=changes(a_ten[1d]),steps=100-16 10.5ms ± 0% 11.8ms ± 1% +12.50% (p=0.000 n=8+10) RangeQuery/expr=changes(a_ten[1d]),steps=1000-16 77.6ms ± 1% 87.4ms ± 1% +12.63% (p=0.000 n=9+9) RangeQuery/expr=changes(a_hundred[1d]),steps=1-16 30.4ms ± 2% 32.8ms ± 1% +8.01% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=10-16 37.1ms ± 2% 40.6ms ± 2% +9.64% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=100-16 105ms ± 1% 117ms ± 1% +11.69% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=1000-16 783ms ± 3% 876ms ± 1% +11.83% (p=0.000 n=9+10) ``` And then runtime v2.39 compared to after this commit: ``` name old time/op new time/op delta RangeQuery/expr=changes(a_one[1d]),steps=1-16 391µs ± 2% 547µs ± 1% +39.84% (p=0.000 n=9+8) RangeQuery/expr=changes(a_one[1d]),steps=10-16 452µs ± 2% 616µs ± 2% +36.15% (p=0.000 n=10+10) RangeQuery/expr=changes(a_one[1d]),steps=100-16 1.12ms ± 1% 1.26ms ± 1% +12.20% (p=0.000 n=8+10) RangeQuery/expr=changes(a_one[1d]),steps=1000-16 7.83ms ± 1% 7.95ms ± 1% +1.59% (p=0.000 n=10+8) RangeQuery/expr=changes(a_ten[1d]),steps=1-16 2.98ms ± 0% 3.38ms ± 2% +13.49% (p=0.000 n=9+10) RangeQuery/expr=changes(a_ten[1d]),steps=10-16 3.66ms ± 1% 4.02ms ± 1% +9.80% (p=0.000 n=10+9) RangeQuery/expr=changes(a_ten[1d]),steps=100-16 10.5ms ± 0% 10.8ms ± 1% +3.08% (p=0.000 n=8+10) RangeQuery/expr=changes(a_ten[1d]),steps=1000-16 77.6ms ± 1% 78.1ms ± 1% +0.58% (p=0.035 n=9+10) RangeQuery/expr=changes(a_hundred[1d]),steps=1-16 30.4ms ± 2% 33.5ms ± 4% +10.18% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=10-16 37.1ms ± 2% 40.0ms ± 1% +7.98% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=100-16 105ms ± 1% 107ms ± 1% +1.92% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=1000-16 783ms ± 3% 775ms ± 1% -1.02% (p=0.019 n=9+9) ``` In summary, the runtime doesn't really improve with this change for queries with just a few steps. For queries with many steps, this commit essentially reinstates the old performance. This is good because the many-step queries are the one that matter most (longest absolute runtime). In terms of allocations, though, this commit doesn't make a dent at all (numbers not shown). The reason is that most of the allocations happen in the sampleRingIterator (in the storage package), which has to be addressed in a separate commit. Signed-off-by: beorn7 <beorn@grafana.com>
2 years ago
queryResult: promql.Vector{{T: 0, F: 1.5}},
},
{
// Get value from query.
text: "{{ query \"metric{instance='a'}\" | first | value }}",
queryResult: promql.Vector{
{
Metric: labels.FromStrings(labels.MetricName, "metric", "instance", "a"),
promql: Separate `Point` into `FPoint` and `HPoint` In other words: Instead of having a “polymorphous” `Point` that can either contain a float value or a histogram value, use an `FPoint` for floats and an `HPoint` for histograms. This seemingly small change has a _lot_ of repercussions throughout the codebase. The idea here is to avoid the increase in size of `Point` arrays that happened after native histograms had been added. The higher-level data structures (`Sample`, `Series`, etc.) are still “polymorphous”. The same idea could be applied to them, but at each step the trade-offs needed to be evaluated. The idea with this change is to do the minimum necessary to get back to pre-histogram performance for functions that do not touch histograms. Here are comparisons for the `changes` function. The test data doesn't include histograms yet. Ideally, there would be no change in the benchmark result at all. First runtime v2.39 compared to directly prior to this commit: ``` name old time/op new time/op delta RangeQuery/expr=changes(a_one[1d]),steps=1-16 391µs ± 2% 542µs ± 1% +38.58% (p=0.000 n=9+8) RangeQuery/expr=changes(a_one[1d]),steps=10-16 452µs ± 2% 617µs ± 2% +36.48% (p=0.000 n=10+10) RangeQuery/expr=changes(a_one[1d]),steps=100-16 1.12ms ± 1% 1.36ms ± 2% +21.58% (p=0.000 n=8+10) RangeQuery/expr=changes(a_one[1d]),steps=1000-16 7.83ms ± 1% 8.94ms ± 1% +14.21% (p=0.000 n=10+10) RangeQuery/expr=changes(a_ten[1d]),steps=1-16 2.98ms ± 0% 3.30ms ± 1% +10.67% (p=0.000 n=9+10) RangeQuery/expr=changes(a_ten[1d]),steps=10-16 3.66ms ± 1% 4.10ms ± 1% +11.82% (p=0.000 n=10+10) RangeQuery/expr=changes(a_ten[1d]),steps=100-16 10.5ms ± 0% 11.8ms ± 1% +12.50% (p=0.000 n=8+10) RangeQuery/expr=changes(a_ten[1d]),steps=1000-16 77.6ms ± 1% 87.4ms ± 1% +12.63% (p=0.000 n=9+9) RangeQuery/expr=changes(a_hundred[1d]),steps=1-16 30.4ms ± 2% 32.8ms ± 1% +8.01% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=10-16 37.1ms ± 2% 40.6ms ± 2% +9.64% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=100-16 105ms ± 1% 117ms ± 1% +11.69% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=1000-16 783ms ± 3% 876ms ± 1% +11.83% (p=0.000 n=9+10) ``` And then runtime v2.39 compared to after this commit: ``` name old time/op new time/op delta RangeQuery/expr=changes(a_one[1d]),steps=1-16 391µs ± 2% 547µs ± 1% +39.84% (p=0.000 n=9+8) RangeQuery/expr=changes(a_one[1d]),steps=10-16 452µs ± 2% 616µs ± 2% +36.15% (p=0.000 n=10+10) RangeQuery/expr=changes(a_one[1d]),steps=100-16 1.12ms ± 1% 1.26ms ± 1% +12.20% (p=0.000 n=8+10) RangeQuery/expr=changes(a_one[1d]),steps=1000-16 7.83ms ± 1% 7.95ms ± 1% +1.59% (p=0.000 n=10+8) RangeQuery/expr=changes(a_ten[1d]),steps=1-16 2.98ms ± 0% 3.38ms ± 2% +13.49% (p=0.000 n=9+10) RangeQuery/expr=changes(a_ten[1d]),steps=10-16 3.66ms ± 1% 4.02ms ± 1% +9.80% (p=0.000 n=10+9) RangeQuery/expr=changes(a_ten[1d]),steps=100-16 10.5ms ± 0% 10.8ms ± 1% +3.08% (p=0.000 n=8+10) RangeQuery/expr=changes(a_ten[1d]),steps=1000-16 77.6ms ± 1% 78.1ms ± 1% +0.58% (p=0.035 n=9+10) RangeQuery/expr=changes(a_hundred[1d]),steps=1-16 30.4ms ± 2% 33.5ms ± 4% +10.18% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=10-16 37.1ms ± 2% 40.0ms ± 1% +7.98% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=100-16 105ms ± 1% 107ms ± 1% +1.92% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=1000-16 783ms ± 3% 775ms ± 1% -1.02% (p=0.019 n=9+9) ``` In summary, the runtime doesn't really improve with this change for queries with just a few steps. For queries with many steps, this commit essentially reinstates the old performance. This is good because the many-step queries are the one that matter most (longest absolute runtime). In terms of allocations, though, this commit doesn't make a dent at all (numbers not shown). The reason is that most of the allocations happen in the sampleRingIterator (in the storage package), which has to be addressed in a separate commit. Signed-off-by: beorn7 <beorn@grafana.com>
2 years ago
T: 0,
F: 11,
},
},
output: "11",
},
{
// Get value of a native histogram from query.
text: "{{ query \"metric{instance='a'}\" | first | value }}",
queryResult: promql.Vector{
{
Metric: labels.FromStrings(labels.MetricName, "metric", "instance", "a"),
T: 0,
H: &histogram.FloatHistogram{Count: 3, Sum: 10},
},
},
output: (&histogram.FloatHistogram{Count: 3, Sum: 10}).String(),
},
{
// Get label from query.
text: "{{ query \"metric{instance='a'}\" | first | label \"instance\" }}",
queryResult: promql.Vector{
{
Metric: labels.FromStrings(labels.MetricName, "metric", "instance", "a"),
promql: Separate `Point` into `FPoint` and `HPoint` In other words: Instead of having a “polymorphous” `Point` that can either contain a float value or a histogram value, use an `FPoint` for floats and an `HPoint` for histograms. This seemingly small change has a _lot_ of repercussions throughout the codebase. The idea here is to avoid the increase in size of `Point` arrays that happened after native histograms had been added. The higher-level data structures (`Sample`, `Series`, etc.) are still “polymorphous”. The same idea could be applied to them, but at each step the trade-offs needed to be evaluated. The idea with this change is to do the minimum necessary to get back to pre-histogram performance for functions that do not touch histograms. Here are comparisons for the `changes` function. The test data doesn't include histograms yet. Ideally, there would be no change in the benchmark result at all. First runtime v2.39 compared to directly prior to this commit: ``` name old time/op new time/op delta RangeQuery/expr=changes(a_one[1d]),steps=1-16 391µs ± 2% 542µs ± 1% +38.58% (p=0.000 n=9+8) RangeQuery/expr=changes(a_one[1d]),steps=10-16 452µs ± 2% 617µs ± 2% +36.48% (p=0.000 n=10+10) RangeQuery/expr=changes(a_one[1d]),steps=100-16 1.12ms ± 1% 1.36ms ± 2% +21.58% (p=0.000 n=8+10) RangeQuery/expr=changes(a_one[1d]),steps=1000-16 7.83ms ± 1% 8.94ms ± 1% +14.21% (p=0.000 n=10+10) RangeQuery/expr=changes(a_ten[1d]),steps=1-16 2.98ms ± 0% 3.30ms ± 1% +10.67% (p=0.000 n=9+10) RangeQuery/expr=changes(a_ten[1d]),steps=10-16 3.66ms ± 1% 4.10ms ± 1% +11.82% (p=0.000 n=10+10) RangeQuery/expr=changes(a_ten[1d]),steps=100-16 10.5ms ± 0% 11.8ms ± 1% +12.50% (p=0.000 n=8+10) RangeQuery/expr=changes(a_ten[1d]),steps=1000-16 77.6ms ± 1% 87.4ms ± 1% +12.63% (p=0.000 n=9+9) RangeQuery/expr=changes(a_hundred[1d]),steps=1-16 30.4ms ± 2% 32.8ms ± 1% +8.01% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=10-16 37.1ms ± 2% 40.6ms ± 2% +9.64% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=100-16 105ms ± 1% 117ms ± 1% +11.69% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=1000-16 783ms ± 3% 876ms ± 1% +11.83% (p=0.000 n=9+10) ``` And then runtime v2.39 compared to after this commit: ``` name old time/op new time/op delta RangeQuery/expr=changes(a_one[1d]),steps=1-16 391µs ± 2% 547µs ± 1% +39.84% (p=0.000 n=9+8) RangeQuery/expr=changes(a_one[1d]),steps=10-16 452µs ± 2% 616µs ± 2% +36.15% (p=0.000 n=10+10) RangeQuery/expr=changes(a_one[1d]),steps=100-16 1.12ms ± 1% 1.26ms ± 1% +12.20% (p=0.000 n=8+10) RangeQuery/expr=changes(a_one[1d]),steps=1000-16 7.83ms ± 1% 7.95ms ± 1% +1.59% (p=0.000 n=10+8) RangeQuery/expr=changes(a_ten[1d]),steps=1-16 2.98ms ± 0% 3.38ms ± 2% +13.49% (p=0.000 n=9+10) RangeQuery/expr=changes(a_ten[1d]),steps=10-16 3.66ms ± 1% 4.02ms ± 1% +9.80% (p=0.000 n=10+9) RangeQuery/expr=changes(a_ten[1d]),steps=100-16 10.5ms ± 0% 10.8ms ± 1% +3.08% (p=0.000 n=8+10) RangeQuery/expr=changes(a_ten[1d]),steps=1000-16 77.6ms ± 1% 78.1ms ± 1% +0.58% (p=0.035 n=9+10) RangeQuery/expr=changes(a_hundred[1d]),steps=1-16 30.4ms ± 2% 33.5ms ± 4% +10.18% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=10-16 37.1ms ± 2% 40.0ms ± 1% +7.98% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=100-16 105ms ± 1% 107ms ± 1% +1.92% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=1000-16 783ms ± 3% 775ms ± 1% -1.02% (p=0.019 n=9+9) ``` In summary, the runtime doesn't really improve with this change for queries with just a few steps. For queries with many steps, this commit essentially reinstates the old performance. This is good because the many-step queries are the one that matter most (longest absolute runtime). In terms of allocations, though, this commit doesn't make a dent at all (numbers not shown). The reason is that most of the allocations happen in the sampleRingIterator (in the storage package), which has to be addressed in a separate commit. Signed-off-by: beorn7 <beorn@grafana.com>
2 years ago
T: 0,
F: 11,
},
},
output: "a",
},
{
// Get label "__value__" from query.
text: "{{ query \"metric{__value__='a'}\" | first | strvalue }}",
queryResult: promql.Vector{
{
Metric: labels.FromStrings(labels.MetricName, "metric", "__value__", "a"),
promql: Separate `Point` into `FPoint` and `HPoint` In other words: Instead of having a “polymorphous” `Point` that can either contain a float value or a histogram value, use an `FPoint` for floats and an `HPoint` for histograms. This seemingly small change has a _lot_ of repercussions throughout the codebase. The idea here is to avoid the increase in size of `Point` arrays that happened after native histograms had been added. The higher-level data structures (`Sample`, `Series`, etc.) are still “polymorphous”. The same idea could be applied to them, but at each step the trade-offs needed to be evaluated. The idea with this change is to do the minimum necessary to get back to pre-histogram performance for functions that do not touch histograms. Here are comparisons for the `changes` function. The test data doesn't include histograms yet. Ideally, there would be no change in the benchmark result at all. First runtime v2.39 compared to directly prior to this commit: ``` name old time/op new time/op delta RangeQuery/expr=changes(a_one[1d]),steps=1-16 391µs ± 2% 542µs ± 1% +38.58% (p=0.000 n=9+8) RangeQuery/expr=changes(a_one[1d]),steps=10-16 452µs ± 2% 617µs ± 2% +36.48% (p=0.000 n=10+10) RangeQuery/expr=changes(a_one[1d]),steps=100-16 1.12ms ± 1% 1.36ms ± 2% +21.58% (p=0.000 n=8+10) RangeQuery/expr=changes(a_one[1d]),steps=1000-16 7.83ms ± 1% 8.94ms ± 1% +14.21% (p=0.000 n=10+10) RangeQuery/expr=changes(a_ten[1d]),steps=1-16 2.98ms ± 0% 3.30ms ± 1% +10.67% (p=0.000 n=9+10) RangeQuery/expr=changes(a_ten[1d]),steps=10-16 3.66ms ± 1% 4.10ms ± 1% +11.82% (p=0.000 n=10+10) RangeQuery/expr=changes(a_ten[1d]),steps=100-16 10.5ms ± 0% 11.8ms ± 1% +12.50% (p=0.000 n=8+10) RangeQuery/expr=changes(a_ten[1d]),steps=1000-16 77.6ms ± 1% 87.4ms ± 1% +12.63% (p=0.000 n=9+9) RangeQuery/expr=changes(a_hundred[1d]),steps=1-16 30.4ms ± 2% 32.8ms ± 1% +8.01% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=10-16 37.1ms ± 2% 40.6ms ± 2% +9.64% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=100-16 105ms ± 1% 117ms ± 1% +11.69% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=1000-16 783ms ± 3% 876ms ± 1% +11.83% (p=0.000 n=9+10) ``` And then runtime v2.39 compared to after this commit: ``` name old time/op new time/op delta RangeQuery/expr=changes(a_one[1d]),steps=1-16 391µs ± 2% 547µs ± 1% +39.84% (p=0.000 n=9+8) RangeQuery/expr=changes(a_one[1d]),steps=10-16 452µs ± 2% 616µs ± 2% +36.15% (p=0.000 n=10+10) RangeQuery/expr=changes(a_one[1d]),steps=100-16 1.12ms ± 1% 1.26ms ± 1% +12.20% (p=0.000 n=8+10) RangeQuery/expr=changes(a_one[1d]),steps=1000-16 7.83ms ± 1% 7.95ms ± 1% +1.59% (p=0.000 n=10+8) RangeQuery/expr=changes(a_ten[1d]),steps=1-16 2.98ms ± 0% 3.38ms ± 2% +13.49% (p=0.000 n=9+10) RangeQuery/expr=changes(a_ten[1d]),steps=10-16 3.66ms ± 1% 4.02ms ± 1% +9.80% (p=0.000 n=10+9) RangeQuery/expr=changes(a_ten[1d]),steps=100-16 10.5ms ± 0% 10.8ms ± 1% +3.08% (p=0.000 n=8+10) RangeQuery/expr=changes(a_ten[1d]),steps=1000-16 77.6ms ± 1% 78.1ms ± 1% +0.58% (p=0.035 n=9+10) RangeQuery/expr=changes(a_hundred[1d]),steps=1-16 30.4ms ± 2% 33.5ms ± 4% +10.18% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=10-16 37.1ms ± 2% 40.0ms ± 1% +7.98% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=100-16 105ms ± 1% 107ms ± 1% +1.92% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=1000-16 783ms ± 3% 775ms ± 1% -1.02% (p=0.019 n=9+9) ``` In summary, the runtime doesn't really improve with this change for queries with just a few steps. For queries with many steps, this commit essentially reinstates the old performance. This is good because the many-step queries are the one that matter most (longest absolute runtime). In terms of allocations, though, this commit doesn't make a dent at all (numbers not shown). The reason is that most of the allocations happen in the sampleRingIterator (in the storage package), which has to be addressed in a separate commit. Signed-off-by: beorn7 <beorn@grafana.com>
2 years ago
T: 0,
F: 11,
},
},
output: "a",
},
{
// Missing label is empty when using label function.
text: "{{ query \"metric{instance='a'}\" | first | label \"foo\" }}",
queryResult: promql.Vector{
{
Metric: labels.FromStrings(labels.MetricName, "metric", "instance", "a"),
promql: Separate `Point` into `FPoint` and `HPoint` In other words: Instead of having a “polymorphous” `Point` that can either contain a float value or a histogram value, use an `FPoint` for floats and an `HPoint` for histograms. This seemingly small change has a _lot_ of repercussions throughout the codebase. The idea here is to avoid the increase in size of `Point` arrays that happened after native histograms had been added. The higher-level data structures (`Sample`, `Series`, etc.) are still “polymorphous”. The same idea could be applied to them, but at each step the trade-offs needed to be evaluated. The idea with this change is to do the minimum necessary to get back to pre-histogram performance for functions that do not touch histograms. Here are comparisons for the `changes` function. The test data doesn't include histograms yet. Ideally, there would be no change in the benchmark result at all. First runtime v2.39 compared to directly prior to this commit: ``` name old time/op new time/op delta RangeQuery/expr=changes(a_one[1d]),steps=1-16 391µs ± 2% 542µs ± 1% +38.58% (p=0.000 n=9+8) RangeQuery/expr=changes(a_one[1d]),steps=10-16 452µs ± 2% 617µs ± 2% +36.48% (p=0.000 n=10+10) RangeQuery/expr=changes(a_one[1d]),steps=100-16 1.12ms ± 1% 1.36ms ± 2% +21.58% (p=0.000 n=8+10) RangeQuery/expr=changes(a_one[1d]),steps=1000-16 7.83ms ± 1% 8.94ms ± 1% +14.21% (p=0.000 n=10+10) RangeQuery/expr=changes(a_ten[1d]),steps=1-16 2.98ms ± 0% 3.30ms ± 1% +10.67% (p=0.000 n=9+10) RangeQuery/expr=changes(a_ten[1d]),steps=10-16 3.66ms ± 1% 4.10ms ± 1% +11.82% (p=0.000 n=10+10) RangeQuery/expr=changes(a_ten[1d]),steps=100-16 10.5ms ± 0% 11.8ms ± 1% +12.50% (p=0.000 n=8+10) RangeQuery/expr=changes(a_ten[1d]),steps=1000-16 77.6ms ± 1% 87.4ms ± 1% +12.63% (p=0.000 n=9+9) RangeQuery/expr=changes(a_hundred[1d]),steps=1-16 30.4ms ± 2% 32.8ms ± 1% +8.01% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=10-16 37.1ms ± 2% 40.6ms ± 2% +9.64% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=100-16 105ms ± 1% 117ms ± 1% +11.69% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=1000-16 783ms ± 3% 876ms ± 1% +11.83% (p=0.000 n=9+10) ``` And then runtime v2.39 compared to after this commit: ``` name old time/op new time/op delta RangeQuery/expr=changes(a_one[1d]),steps=1-16 391µs ± 2% 547µs ± 1% +39.84% (p=0.000 n=9+8) RangeQuery/expr=changes(a_one[1d]),steps=10-16 452µs ± 2% 616µs ± 2% +36.15% (p=0.000 n=10+10) RangeQuery/expr=changes(a_one[1d]),steps=100-16 1.12ms ± 1% 1.26ms ± 1% +12.20% (p=0.000 n=8+10) RangeQuery/expr=changes(a_one[1d]),steps=1000-16 7.83ms ± 1% 7.95ms ± 1% +1.59% (p=0.000 n=10+8) RangeQuery/expr=changes(a_ten[1d]),steps=1-16 2.98ms ± 0% 3.38ms ± 2% +13.49% (p=0.000 n=9+10) RangeQuery/expr=changes(a_ten[1d]),steps=10-16 3.66ms ± 1% 4.02ms ± 1% +9.80% (p=0.000 n=10+9) RangeQuery/expr=changes(a_ten[1d]),steps=100-16 10.5ms ± 0% 10.8ms ± 1% +3.08% (p=0.000 n=8+10) RangeQuery/expr=changes(a_ten[1d]),steps=1000-16 77.6ms ± 1% 78.1ms ± 1% +0.58% (p=0.035 n=9+10) RangeQuery/expr=changes(a_hundred[1d]),steps=1-16 30.4ms ± 2% 33.5ms ± 4% +10.18% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=10-16 37.1ms ± 2% 40.0ms ± 1% +7.98% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=100-16 105ms ± 1% 107ms ± 1% +1.92% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=1000-16 783ms ± 3% 775ms ± 1% -1.02% (p=0.019 n=9+9) ``` In summary, the runtime doesn't really improve with this change for queries with just a few steps. For queries with many steps, this commit essentially reinstates the old performance. This is good because the many-step queries are the one that matter most (longest absolute runtime). In terms of allocations, though, this commit doesn't make a dent at all (numbers not shown). The reason is that most of the allocations happen in the sampleRingIterator (in the storage package), which has to be addressed in a separate commit. Signed-off-by: beorn7 <beorn@grafana.com>
2 years ago
T: 0,
F: 11,
},
},
output: "",
},
{
// Missing label is empty when not using label function.
text: "{{ $x := query \"metric\" | first }}{{ $x.Labels.foo }}",
queryResult: promql.Vector{
{
Metric: labels.FromStrings(labels.MetricName, "metric", "instance", "a"),
promql: Separate `Point` into `FPoint` and `HPoint` In other words: Instead of having a “polymorphous” `Point` that can either contain a float value or a histogram value, use an `FPoint` for floats and an `HPoint` for histograms. This seemingly small change has a _lot_ of repercussions throughout the codebase. The idea here is to avoid the increase in size of `Point` arrays that happened after native histograms had been added. The higher-level data structures (`Sample`, `Series`, etc.) are still “polymorphous”. The same idea could be applied to them, but at each step the trade-offs needed to be evaluated. The idea with this change is to do the minimum necessary to get back to pre-histogram performance for functions that do not touch histograms. Here are comparisons for the `changes` function. The test data doesn't include histograms yet. Ideally, there would be no change in the benchmark result at all. First runtime v2.39 compared to directly prior to this commit: ``` name old time/op new time/op delta RangeQuery/expr=changes(a_one[1d]),steps=1-16 391µs ± 2% 542µs ± 1% +38.58% (p=0.000 n=9+8) RangeQuery/expr=changes(a_one[1d]),steps=10-16 452µs ± 2% 617µs ± 2% +36.48% (p=0.000 n=10+10) RangeQuery/expr=changes(a_one[1d]),steps=100-16 1.12ms ± 1% 1.36ms ± 2% +21.58% (p=0.000 n=8+10) RangeQuery/expr=changes(a_one[1d]),steps=1000-16 7.83ms ± 1% 8.94ms ± 1% +14.21% (p=0.000 n=10+10) RangeQuery/expr=changes(a_ten[1d]),steps=1-16 2.98ms ± 0% 3.30ms ± 1% +10.67% (p=0.000 n=9+10) RangeQuery/expr=changes(a_ten[1d]),steps=10-16 3.66ms ± 1% 4.10ms ± 1% +11.82% (p=0.000 n=10+10) RangeQuery/expr=changes(a_ten[1d]),steps=100-16 10.5ms ± 0% 11.8ms ± 1% +12.50% (p=0.000 n=8+10) RangeQuery/expr=changes(a_ten[1d]),steps=1000-16 77.6ms ± 1% 87.4ms ± 1% +12.63% (p=0.000 n=9+9) RangeQuery/expr=changes(a_hundred[1d]),steps=1-16 30.4ms ± 2% 32.8ms ± 1% +8.01% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=10-16 37.1ms ± 2% 40.6ms ± 2% +9.64% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=100-16 105ms ± 1% 117ms ± 1% +11.69% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=1000-16 783ms ± 3% 876ms ± 1% +11.83% (p=0.000 n=9+10) ``` And then runtime v2.39 compared to after this commit: ``` name old time/op new time/op delta RangeQuery/expr=changes(a_one[1d]),steps=1-16 391µs ± 2% 547µs ± 1% +39.84% (p=0.000 n=9+8) RangeQuery/expr=changes(a_one[1d]),steps=10-16 452µs ± 2% 616µs ± 2% +36.15% (p=0.000 n=10+10) RangeQuery/expr=changes(a_one[1d]),steps=100-16 1.12ms ± 1% 1.26ms ± 1% +12.20% (p=0.000 n=8+10) RangeQuery/expr=changes(a_one[1d]),steps=1000-16 7.83ms ± 1% 7.95ms ± 1% +1.59% (p=0.000 n=10+8) RangeQuery/expr=changes(a_ten[1d]),steps=1-16 2.98ms ± 0% 3.38ms ± 2% +13.49% (p=0.000 n=9+10) RangeQuery/expr=changes(a_ten[1d]),steps=10-16 3.66ms ± 1% 4.02ms ± 1% +9.80% (p=0.000 n=10+9) RangeQuery/expr=changes(a_ten[1d]),steps=100-16 10.5ms ± 0% 10.8ms ± 1% +3.08% (p=0.000 n=8+10) RangeQuery/expr=changes(a_ten[1d]),steps=1000-16 77.6ms ± 1% 78.1ms ± 1% +0.58% (p=0.035 n=9+10) RangeQuery/expr=changes(a_hundred[1d]),steps=1-16 30.4ms ± 2% 33.5ms ± 4% +10.18% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=10-16 37.1ms ± 2% 40.0ms ± 1% +7.98% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=100-16 105ms ± 1% 107ms ± 1% +1.92% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=1000-16 783ms ± 3% 775ms ± 1% -1.02% (p=0.019 n=9+9) ``` In summary, the runtime doesn't really improve with this change for queries with just a few steps. For queries with many steps, this commit essentially reinstates the old performance. This is good because the many-step queries are the one that matter most (longest absolute runtime). In terms of allocations, though, this commit doesn't make a dent at all (numbers not shown). The reason is that most of the allocations happen in the sampleRingIterator (in the storage package), which has to be addressed in a separate commit. Signed-off-by: beorn7 <beorn@grafana.com>
2 years ago
T: 0,
F: 11,
},
},
output: "",
},
{
text: "{{ $x := query \"metric\" | first }}{{ $x.Labels.foo }}",
queryResult: promql.Vector{
{
Metric: labels.FromStrings(labels.MetricName, "metric", "instance", "a"),
promql: Separate `Point` into `FPoint` and `HPoint` In other words: Instead of having a “polymorphous” `Point` that can either contain a float value or a histogram value, use an `FPoint` for floats and an `HPoint` for histograms. This seemingly small change has a _lot_ of repercussions throughout the codebase. The idea here is to avoid the increase in size of `Point` arrays that happened after native histograms had been added. The higher-level data structures (`Sample`, `Series`, etc.) are still “polymorphous”. The same idea could be applied to them, but at each step the trade-offs needed to be evaluated. The idea with this change is to do the minimum necessary to get back to pre-histogram performance for functions that do not touch histograms. Here are comparisons for the `changes` function. The test data doesn't include histograms yet. Ideally, there would be no change in the benchmark result at all. First runtime v2.39 compared to directly prior to this commit: ``` name old time/op new time/op delta RangeQuery/expr=changes(a_one[1d]),steps=1-16 391µs ± 2% 542µs ± 1% +38.58% (p=0.000 n=9+8) RangeQuery/expr=changes(a_one[1d]),steps=10-16 452µs ± 2% 617µs ± 2% +36.48% (p=0.000 n=10+10) RangeQuery/expr=changes(a_one[1d]),steps=100-16 1.12ms ± 1% 1.36ms ± 2% +21.58% (p=0.000 n=8+10) RangeQuery/expr=changes(a_one[1d]),steps=1000-16 7.83ms ± 1% 8.94ms ± 1% +14.21% (p=0.000 n=10+10) RangeQuery/expr=changes(a_ten[1d]),steps=1-16 2.98ms ± 0% 3.30ms ± 1% +10.67% (p=0.000 n=9+10) RangeQuery/expr=changes(a_ten[1d]),steps=10-16 3.66ms ± 1% 4.10ms ± 1% +11.82% (p=0.000 n=10+10) RangeQuery/expr=changes(a_ten[1d]),steps=100-16 10.5ms ± 0% 11.8ms ± 1% +12.50% (p=0.000 n=8+10) RangeQuery/expr=changes(a_ten[1d]),steps=1000-16 77.6ms ± 1% 87.4ms ± 1% +12.63% (p=0.000 n=9+9) RangeQuery/expr=changes(a_hundred[1d]),steps=1-16 30.4ms ± 2% 32.8ms ± 1% +8.01% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=10-16 37.1ms ± 2% 40.6ms ± 2% +9.64% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=100-16 105ms ± 1% 117ms ± 1% +11.69% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=1000-16 783ms ± 3% 876ms ± 1% +11.83% (p=0.000 n=9+10) ``` And then runtime v2.39 compared to after this commit: ``` name old time/op new time/op delta RangeQuery/expr=changes(a_one[1d]),steps=1-16 391µs ± 2% 547µs ± 1% +39.84% (p=0.000 n=9+8) RangeQuery/expr=changes(a_one[1d]),steps=10-16 452µs ± 2% 616µs ± 2% +36.15% (p=0.000 n=10+10) RangeQuery/expr=changes(a_one[1d]),steps=100-16 1.12ms ± 1% 1.26ms ± 1% +12.20% (p=0.000 n=8+10) RangeQuery/expr=changes(a_one[1d]),steps=1000-16 7.83ms ± 1% 7.95ms ± 1% +1.59% (p=0.000 n=10+8) RangeQuery/expr=changes(a_ten[1d]),steps=1-16 2.98ms ± 0% 3.38ms ± 2% +13.49% (p=0.000 n=9+10) RangeQuery/expr=changes(a_ten[1d]),steps=10-16 3.66ms ± 1% 4.02ms ± 1% +9.80% (p=0.000 n=10+9) RangeQuery/expr=changes(a_ten[1d]),steps=100-16 10.5ms ± 0% 10.8ms ± 1% +3.08% (p=0.000 n=8+10) RangeQuery/expr=changes(a_ten[1d]),steps=1000-16 77.6ms ± 1% 78.1ms ± 1% +0.58% (p=0.035 n=9+10) RangeQuery/expr=changes(a_hundred[1d]),steps=1-16 30.4ms ± 2% 33.5ms ± 4% +10.18% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=10-16 37.1ms ± 2% 40.0ms ± 1% +7.98% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=100-16 105ms ± 1% 107ms ± 1% +1.92% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=1000-16 783ms ± 3% 775ms ± 1% -1.02% (p=0.019 n=9+9) ``` In summary, the runtime doesn't really improve with this change for queries with just a few steps. For queries with many steps, this commit essentially reinstates the old performance. This is good because the many-step queries are the one that matter most (longest absolute runtime). In terms of allocations, though, this commit doesn't make a dent at all (numbers not shown). The reason is that most of the allocations happen in the sampleRingIterator (in the storage package), which has to be addressed in a separate commit. Signed-off-by: beorn7 <beorn@grafana.com>
2 years ago
T: 0,
F: 11,
},
},
output: "",
html: true,
},
{
// Range over query and sort by label.
text: "{{ range query \"metric\" | sortByLabel \"instance\" }}{{.Labels.instance}}:{{.Value}}: {{end}}",
queryResult: promql.Vector{
{
Metric: labels.FromStrings(labels.MetricName, "metric", "instance", "b"),
promql: Separate `Point` into `FPoint` and `HPoint` In other words: Instead of having a “polymorphous” `Point` that can either contain a float value or a histogram value, use an `FPoint` for floats and an `HPoint` for histograms. This seemingly small change has a _lot_ of repercussions throughout the codebase. The idea here is to avoid the increase in size of `Point` arrays that happened after native histograms had been added. The higher-level data structures (`Sample`, `Series`, etc.) are still “polymorphous”. The same idea could be applied to them, but at each step the trade-offs needed to be evaluated. The idea with this change is to do the minimum necessary to get back to pre-histogram performance for functions that do not touch histograms. Here are comparisons for the `changes` function. The test data doesn't include histograms yet. Ideally, there would be no change in the benchmark result at all. First runtime v2.39 compared to directly prior to this commit: ``` name old time/op new time/op delta RangeQuery/expr=changes(a_one[1d]),steps=1-16 391µs ± 2% 542µs ± 1% +38.58% (p=0.000 n=9+8) RangeQuery/expr=changes(a_one[1d]),steps=10-16 452µs ± 2% 617µs ± 2% +36.48% (p=0.000 n=10+10) RangeQuery/expr=changes(a_one[1d]),steps=100-16 1.12ms ± 1% 1.36ms ± 2% +21.58% (p=0.000 n=8+10) RangeQuery/expr=changes(a_one[1d]),steps=1000-16 7.83ms ± 1% 8.94ms ± 1% +14.21% (p=0.000 n=10+10) RangeQuery/expr=changes(a_ten[1d]),steps=1-16 2.98ms ± 0% 3.30ms ± 1% +10.67% (p=0.000 n=9+10) RangeQuery/expr=changes(a_ten[1d]),steps=10-16 3.66ms ± 1% 4.10ms ± 1% +11.82% (p=0.000 n=10+10) RangeQuery/expr=changes(a_ten[1d]),steps=100-16 10.5ms ± 0% 11.8ms ± 1% +12.50% (p=0.000 n=8+10) RangeQuery/expr=changes(a_ten[1d]),steps=1000-16 77.6ms ± 1% 87.4ms ± 1% +12.63% (p=0.000 n=9+9) RangeQuery/expr=changes(a_hundred[1d]),steps=1-16 30.4ms ± 2% 32.8ms ± 1% +8.01% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=10-16 37.1ms ± 2% 40.6ms ± 2% +9.64% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=100-16 105ms ± 1% 117ms ± 1% +11.69% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=1000-16 783ms ± 3% 876ms ± 1% +11.83% (p=0.000 n=9+10) ``` And then runtime v2.39 compared to after this commit: ``` name old time/op new time/op delta RangeQuery/expr=changes(a_one[1d]),steps=1-16 391µs ± 2% 547µs ± 1% +39.84% (p=0.000 n=9+8) RangeQuery/expr=changes(a_one[1d]),steps=10-16 452µs ± 2% 616µs ± 2% +36.15% (p=0.000 n=10+10) RangeQuery/expr=changes(a_one[1d]),steps=100-16 1.12ms ± 1% 1.26ms ± 1% +12.20% (p=0.000 n=8+10) RangeQuery/expr=changes(a_one[1d]),steps=1000-16 7.83ms ± 1% 7.95ms ± 1% +1.59% (p=0.000 n=10+8) RangeQuery/expr=changes(a_ten[1d]),steps=1-16 2.98ms ± 0% 3.38ms ± 2% +13.49% (p=0.000 n=9+10) RangeQuery/expr=changes(a_ten[1d]),steps=10-16 3.66ms ± 1% 4.02ms ± 1% +9.80% (p=0.000 n=10+9) RangeQuery/expr=changes(a_ten[1d]),steps=100-16 10.5ms ± 0% 10.8ms ± 1% +3.08% (p=0.000 n=8+10) RangeQuery/expr=changes(a_ten[1d]),steps=1000-16 77.6ms ± 1% 78.1ms ± 1% +0.58% (p=0.035 n=9+10) RangeQuery/expr=changes(a_hundred[1d]),steps=1-16 30.4ms ± 2% 33.5ms ± 4% +10.18% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=10-16 37.1ms ± 2% 40.0ms ± 1% +7.98% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=100-16 105ms ± 1% 107ms ± 1% +1.92% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=1000-16 783ms ± 3% 775ms ± 1% -1.02% (p=0.019 n=9+9) ``` In summary, the runtime doesn't really improve with this change for queries with just a few steps. For queries with many steps, this commit essentially reinstates the old performance. This is good because the many-step queries are the one that matter most (longest absolute runtime). In terms of allocations, though, this commit doesn't make a dent at all (numbers not shown). The reason is that most of the allocations happen in the sampleRingIterator (in the storage package), which has to be addressed in a separate commit. Signed-off-by: beorn7 <beorn@grafana.com>
2 years ago
T: 0,
F: 21,
}, {
Metric: labels.FromStrings(labels.MetricName, "metric", "instance", "a"),
promql: Separate `Point` into `FPoint` and `HPoint` In other words: Instead of having a “polymorphous” `Point` that can either contain a float value or a histogram value, use an `FPoint` for floats and an `HPoint` for histograms. This seemingly small change has a _lot_ of repercussions throughout the codebase. The idea here is to avoid the increase in size of `Point` arrays that happened after native histograms had been added. The higher-level data structures (`Sample`, `Series`, etc.) are still “polymorphous”. The same idea could be applied to them, but at each step the trade-offs needed to be evaluated. The idea with this change is to do the minimum necessary to get back to pre-histogram performance for functions that do not touch histograms. Here are comparisons for the `changes` function. The test data doesn't include histograms yet. Ideally, there would be no change in the benchmark result at all. First runtime v2.39 compared to directly prior to this commit: ``` name old time/op new time/op delta RangeQuery/expr=changes(a_one[1d]),steps=1-16 391µs ± 2% 542µs ± 1% +38.58% (p=0.000 n=9+8) RangeQuery/expr=changes(a_one[1d]),steps=10-16 452µs ± 2% 617µs ± 2% +36.48% (p=0.000 n=10+10) RangeQuery/expr=changes(a_one[1d]),steps=100-16 1.12ms ± 1% 1.36ms ± 2% +21.58% (p=0.000 n=8+10) RangeQuery/expr=changes(a_one[1d]),steps=1000-16 7.83ms ± 1% 8.94ms ± 1% +14.21% (p=0.000 n=10+10) RangeQuery/expr=changes(a_ten[1d]),steps=1-16 2.98ms ± 0% 3.30ms ± 1% +10.67% (p=0.000 n=9+10) RangeQuery/expr=changes(a_ten[1d]),steps=10-16 3.66ms ± 1% 4.10ms ± 1% +11.82% (p=0.000 n=10+10) RangeQuery/expr=changes(a_ten[1d]),steps=100-16 10.5ms ± 0% 11.8ms ± 1% +12.50% (p=0.000 n=8+10) RangeQuery/expr=changes(a_ten[1d]),steps=1000-16 77.6ms ± 1% 87.4ms ± 1% +12.63% (p=0.000 n=9+9) RangeQuery/expr=changes(a_hundred[1d]),steps=1-16 30.4ms ± 2% 32.8ms ± 1% +8.01% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=10-16 37.1ms ± 2% 40.6ms ± 2% +9.64% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=100-16 105ms ± 1% 117ms ± 1% +11.69% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=1000-16 783ms ± 3% 876ms ± 1% +11.83% (p=0.000 n=9+10) ``` And then runtime v2.39 compared to after this commit: ``` name old time/op new time/op delta RangeQuery/expr=changes(a_one[1d]),steps=1-16 391µs ± 2% 547µs ± 1% +39.84% (p=0.000 n=9+8) RangeQuery/expr=changes(a_one[1d]),steps=10-16 452µs ± 2% 616µs ± 2% +36.15% (p=0.000 n=10+10) RangeQuery/expr=changes(a_one[1d]),steps=100-16 1.12ms ± 1% 1.26ms ± 1% +12.20% (p=0.000 n=8+10) RangeQuery/expr=changes(a_one[1d]),steps=1000-16 7.83ms ± 1% 7.95ms ± 1% +1.59% (p=0.000 n=10+8) RangeQuery/expr=changes(a_ten[1d]),steps=1-16 2.98ms ± 0% 3.38ms ± 2% +13.49% (p=0.000 n=9+10) RangeQuery/expr=changes(a_ten[1d]),steps=10-16 3.66ms ± 1% 4.02ms ± 1% +9.80% (p=0.000 n=10+9) RangeQuery/expr=changes(a_ten[1d]),steps=100-16 10.5ms ± 0% 10.8ms ± 1% +3.08% (p=0.000 n=8+10) RangeQuery/expr=changes(a_ten[1d]),steps=1000-16 77.6ms ± 1% 78.1ms ± 1% +0.58% (p=0.035 n=9+10) RangeQuery/expr=changes(a_hundred[1d]),steps=1-16 30.4ms ± 2% 33.5ms ± 4% +10.18% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=10-16 37.1ms ± 2% 40.0ms ± 1% +7.98% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=100-16 105ms ± 1% 107ms ± 1% +1.92% (p=0.000 n=10+10) RangeQuery/expr=changes(a_hundred[1d]),steps=1000-16 783ms ± 3% 775ms ± 1% -1.02% (p=0.019 n=9+9) ``` In summary, the runtime doesn't really improve with this change for queries with just a few steps. For queries with many steps, this commit essentially reinstates the old performance. This is good because the many-step queries are the one that matter most (longest absolute runtime). In terms of allocations, though, this commit doesn't make a dent at all (numbers not shown). The reason is that most of the allocations happen in the sampleRingIterator (in the storage package), which has to be addressed in a separate commit. Signed-off-by: beorn7 <beorn@grafana.com>
2 years ago
T: 0,
F: 11,
},
},
output: "a:11: b:21: ",
},
{
// Simple hostname.
text: "{{ \"foo.example.com\" | stripPort }}",
output: "foo.example.com",
},
{
// Hostname with port.
text: "{{ \"foo.example.com:12345\" | stripPort }}",
output: "foo.example.com",
},
{
// Simple IPv4 address.
text: "{{ \"192.0.2.1\" | stripPort }}",
output: "192.0.2.1",
},
{
// IPv4 address with port.
text: "{{ \"192.0.2.1:12345\" | stripPort }}",
output: "192.0.2.1",
},
{
// Simple IPv6 address.
text: "{{ \"2001:0DB8::1\" | stripPort }}",
output: "2001:0DB8::1",
},
{
// IPv6 address with port.
text: "{{ \"[2001:0DB8::1]:12345\" | stripPort }}",
output: "2001:0DB8::1",
},
{
// Value can't be split into host and port.
text: "{{ \"[2001:0DB8::1]::12345\" | stripPort }}",
output: "[2001:0DB8::1]::12345",
},
{
// Missing value is no value for nil options.
text: "{{ .Foo }}",
output: "<no value>",
},
{
// Missing value is no value for no options.
text: "{{ .Foo }}",
options: make([]string, 0),
output: "<no value>",
},
{
// Assert that missing value returns error with missingkey=error.
text: "{{ .Foo }}",
options: []string{"missingkey=error"},
shouldFail: true,
errorMsg: `error executing template test: template: test:1:3: executing "test" at <.Foo>: nil data; no entry for key "Foo"`,
},
{
// Missing value is "" for nil options in ExpandHTML.
text: "{{ .Foo }}",
output: "",
html: true,
},
{
// Missing value is "" for no options in ExpandHTML.
text: "{{ .Foo }}",
options: make([]string, 0),
output: "",
html: true,
},
{
// Assert that missing value returns error with missingkey=error in ExpandHTML.
text: "{{ .Foo }}",
options: []string{"missingkey=error"},
shouldFail: true,
errorMsg: `error executing template test: template: test:1:3: executing "test" at <.Foo>: nil data; no entry for key "Foo"`,
html: true,
},
{
// Unparsable template.
text: "{{",
shouldFail: true,
errorMsg: "error parsing template test: template: test:1: unclosed action",
},
{
// Error in function.
text: "{{ query \"missing\" | first }}",
queryResult: promql.Vector{},
shouldFail: true,
errorMsg: "error executing template test: template: test:1:21: executing \"test\" at <first>: error calling first: first() called on vector with no elements",
},
{
// Panic.
text: "{{ (query \"missing\").banana }}",
queryResult: promql.Vector{},
shouldFail: true,
errorMsg: "error executing template test: template: test:1:10: executing \"test\" at <\"missing\">: can't evaluate field banana in type template.queryResult",
},
{
// Regex replacement.
text: "{{ reReplaceAll \"(a)b\" \"x$1\" \"ab\" }}",
output: "xa",
},
{
// Humanize - float64.
text: "{{ range . }}{{ humanize . }}:{{ end }}",
input: []float64{0.0, 1.0, 1234567.0, .12},
output: "0:1:1.235M:120m:",
},
{
// Humanize - string.
text: "{{ range . }}{{ humanize . }}:{{ end }}",
input: []string{"0.0", "1.0", "1234567.0", ".12"},
output: "0:1:1.235M:120m:",
},
{
// Humanize - string with error.
text: `{{ humanize "one" }}`,
shouldFail: true,
errorMsg: `error executing template test: template: test:1:3: executing "test" at <humanize "one">: error calling humanize: strconv.ParseFloat: parsing "one": invalid syntax`,
},
{
// Humanize - int.
text: "{{ range . }}{{ humanize . }}:{{ end }}",
input: []int64{0, -1, 1, 1234567, math.MaxInt64},
output: "0:-1:1:1.235M:9.223E:",
},
{
// Humanize - uint.
text: "{{ range . }}{{ humanize . }}:{{ end }}",
input: []uint64{0, 1, 1234567, math.MaxUint64},
output: "0:1:1.235M:18.45E:",
},
{
// Humanize1024 - float64.
text: "{{ range . }}{{ humanize1024 . }}:{{ end }}",
input: []float64{0.0, 1.0, 1048576.0, .12},
output: "0:1:1Mi:0.12:",
},
{
// Humanize1024 - string.
text: "{{ range . }}{{ humanize1024 . }}:{{ end }}",
input: []string{"0.0", "1.0", "1048576.0", ".12"},
output: "0:1:1Mi:0.12:",
},
{
// Humanize1024 - string with error.
text: `{{ humanize1024 "one" }}`,
shouldFail: true,
errorMsg: `error executing template test: template: test:1:3: executing "test" at <humanize1024 "one">: error calling humanize1024: strconv.ParseFloat: parsing "one": invalid syntax`,
},
{
// Humanize1024 - int.
text: "{{ range . }}{{ humanize1024 . }}:{{ end }}",
input: []int64{0, -1, 1, 1234567, math.MaxInt64},
output: "0:-1:1:1.177Mi:8Ei:",
},
{
// Humanize1024 - uint.
text: "{{ range . }}{{ humanize1024 . }}:{{ end }}",
input: []uint64{0, 1, 1234567, math.MaxUint64},
output: "0:1:1.177Mi:16Ei:",
},
{
// HumanizeDuration - seconds - float64.
text: "{{ range . }}{{ humanizeDuration . }}:{{ end }}",
input: []float64{0, 1, 60, 3600, 86400, 86400 + 3600, -(86400*2 + 3600*3 + 60*4 + 5), 899.99},
output: "0s:1s:1m 0s:1h 0m 0s:1d 0h 0m 0s:1d 1h 0m 0s:-2d 3h 4m 5s:14m 59s:",
},
{
// HumanizeDuration - seconds - string.
text: "{{ range . }}{{ humanizeDuration . }}:{{ end }}",
input: []string{"0", "1", "60", "3600", "86400"},
output: "0s:1s:1m 0s:1h 0m 0s:1d 0h 0m 0s:",
},
{
// HumanizeDuration - subsecond and fractional seconds - float64.
text: "{{ range . }}{{ humanizeDuration . }}:{{ end }}",
input: []float64{.1, .0001, .12345, 60.1, 60.5, 1.2345, 12.345},
output: "100ms:100us:123.5ms:1m 0s:1m 0s:1.234s:12.35s:",
},
{
// HumanizeDuration - subsecond and fractional seconds - string.
text: "{{ range . }}{{ humanizeDuration . }}:{{ end }}",
input: []string{".1", ".0001", ".12345", "60.1", "60.5", "1.2345", "12.345"},
output: "100ms:100us:123.5ms:1m 0s:1m 0s:1.234s:12.35s:",
},
{
// HumanizeDuration - string with error.
text: `{{ humanizeDuration "one" }}`,
shouldFail: true,
errorMsg: `error executing template test: template: test:1:3: executing "test" at <humanizeDuration "one">: error calling humanizeDuration: strconv.ParseFloat: parsing "one": invalid syntax`,
},
{
// HumanizeDuration - int.
text: "{{ range . }}{{ humanizeDuration . }}:{{ end }}",
input: []int{0, -1, 1, 1234567},
output: "0s:-1s:1s:14d 6h 56m 7s:",
},
{
// HumanizeDuration - uint.
text: "{{ range . }}{{ humanizeDuration . }}:{{ end }}",
input: []uint{0, 1, 1234567},
output: "0s:1s:14d 6h 56m 7s:",
},
{
// Humanize* Inf and NaN - float64.
text: "{{ range . }}{{ humanize . }}:{{ humanize1024 . }}:{{ humanizeDuration . }}:{{humanizeTimestamp .}}:{{ end }}",
input: []float64{math.Inf(1), math.Inf(-1), math.NaN()},
output: "+Inf:+Inf:+Inf:+Inf:-Inf:-Inf:-Inf:-Inf:NaN:NaN:NaN:NaN:",
},
{
// Humanize* Inf and NaN - string.
text: "{{ range . }}{{ humanize . }}:{{ humanize1024 . }}:{{ humanizeDuration . }}:{{humanizeTimestamp .}}:{{ end }}",
input: []string{"+Inf", "-Inf", "NaN"},
output: "+Inf:+Inf:+Inf:+Inf:-Inf:-Inf:-Inf:-Inf:NaN:NaN:NaN:NaN:",
},
{
// HumanizePercentage - model.SampleValue input - float64.
text: "{{ -0.22222 | humanizePercentage }}:{{ 0.0 | humanizePercentage }}:{{ 0.1234567 | humanizePercentage }}:{{ 1.23456 | humanizePercentage }}",
output: "-22.22%:0%:12.35%:123.5%",
},
{
// HumanizePercentage - int.
text: "{{ range . }}{{ humanizePercentage . }}:{{ end }}",
input: []int64{0, -1, 1, 1234567, math.MaxInt64},
output: "0%:-100%:100%:1.235e+08%:9.223e+20%:",
},
{
// HumanizePercentage - uint.
text: "{{ range . }}{{ humanizePercentage . }}:{{ end }}",
input: []uint64{0, 1, 1234567, math.MaxUint64},
output: "0%:100%:1.235e+08%:1.845e+21%:",
},
{
// HumanizePercentage - model.SampleValue input - string.
text: `{{ "-0.22222" | humanizePercentage }}:{{ "0.0" | humanizePercentage }}:{{ "0.1234567" | humanizePercentage }}:{{ "1.23456" | humanizePercentage }}`,
output: "-22.22%:0%:12.35%:123.5%",
},
{
// HumanizePercentage - model.SampleValue input - string with error.
text: `{{ "one" | humanizePercentage }}`,
shouldFail: true,
errorMsg: `error executing template test: template: test:1:11: executing "test" at <humanizePercentage>: error calling humanizePercentage: strconv.ParseFloat: parsing "one": invalid syntax`,
},
{
// HumanizeTimestamp - int.
text: "{{ range . }}{{ humanizeTimestamp . }}:{{ end }}",
input: []int64{0, -1, 1, 1234567, 9223372036},
output: "1970-01-01 00:00:00 +0000 UTC:1969-12-31 23:59:59 +0000 UTC:1970-01-01 00:00:01 +0000 UTC:1970-01-15 06:56:07 +0000 UTC:2262-04-11 23:47:16 +0000 UTC:",
},
{
// HumanizeTimestamp - uint.
text: "{{ range . }}{{ humanizeTimestamp . }}:{{ end }}",
input: []uint64{0, 1, 1234567, 9223372036},
output: "1970-01-01 00:00:00 +0000 UTC:1970-01-01 00:00:01 +0000 UTC:1970-01-15 06:56:07 +0000 UTC:2262-04-11 23:47:16 +0000 UTC:",
},
{
// HumanizeTimestamp - int with error.
text: "{{ range . }}{{ humanizeTimestamp . }}:{{ end }}",
input: []int64{math.MinInt64, math.MaxInt64},
shouldFail: true,
errorMsg: `error executing template test: template: test:1:16: executing "test" at <humanizeTimestamp .>: error calling humanizeTimestamp: -9.223372036854776e+18 cannot be represented as a nanoseconds timestamp since it overflows int64`,
},
{
// HumanizeTimestamp - uint with error.
text: "{{ range . }}{{ humanizeTimestamp . }}:{{ end }}",
input: []uint64{math.MaxUint64},
shouldFail: true,
errorMsg: `error executing template test: template: test:1:16: executing "test" at <humanizeTimestamp .>: error calling humanizeTimestamp: 1.8446744073709552e+19 cannot be represented as a nanoseconds timestamp since it overflows int64`,
},
{
// HumanizeTimestamp - model.SampleValue input - float64.
text: "{{ 1435065584.128 | humanizeTimestamp }}",
output: "2015-06-23 13:19:44.128 +0000 UTC",
},
{
// HumanizeTimestamp - model.SampleValue input - string.
text: `{{ "1435065584.128" | humanizeTimestamp }}`,
output: "2015-06-23 13:19:44.128 +0000 UTC",
},
{
// ToTime - model.SampleValue input - float64.
text: `{{ (1435065584.128 | toTime).Format "2006" }}`,
output: "2015",
},
{
// ToTime - model.SampleValue input - string.
text: `{{ ("1435065584.128" | toTime).Format "2006" }}`,
output: "2015",
},
{
// Title.
text: "{{ \"aa bb CC\" | title }}",
output: "Aa Bb CC",
},
{
// toUpper.
text: "{{ \"aa bb CC\" | toUpper }}",
output: "AA BB CC",
},
{
// toLower.
text: "{{ \"aA bB CC\" | toLower }}",
output: "aa bb cc",
},
{
// Match.
text: "{{ match \"a+\" \"aa\" }} {{ match \"a+\" \"b\" }}",
output: "true false",
},
{
// graphLink.
text: "{{ graphLink \"up\" }}",
output: "/graph?g0.expr=up&g0.tab=0",
},
{
// tableLink.
text: "{{ tableLink \"up\" }}",
output: "/graph?g0.expr=up&g0.tab=1",
},
{
// tmpl.
text: "{{ define \"a\" }}x{{ end }}{{ $name := \"a\"}}{{ tmpl $name . }}",
output: "x",
html: true,
},
{
// pathPrefix.
text: "{{ pathPrefix }}",
output: "/path/prefix",
},
{
// externalURL.
text: "{{ externalURL }}",
output: "http://testhost:9090/path/prefix",
},
{
// parseDuration (using printf to ensure the return is a string).
text: "{{ printf \"%0.2f\" (parseDuration \"1h2m10ms\") }}",
output: "3720.01",
},
{
// Simple hostname.
text: "{{ \"foo.example.com\" | stripDomain }}",
output: "foo",
},
{
// Hostname with port.
text: "{{ \"foo.example.com:12345\" | stripDomain }}",
output: "foo:12345",
},
{
// Simple IPv4 address.
text: "{{ \"192.0.2.1\" | stripDomain }}",
output: "192.0.2.1",
},
{
// IPv4 address with port.
text: "{{ \"192.0.2.1:12345\" | stripDomain }}",
output: "192.0.2.1:12345",
},
{
// Simple IPv6 address.
text: "{{ \"2001:0DB8::1\" | stripDomain }}",
output: "2001:0DB8::1",
},
{
// IPv6 address with port.
text: "{{ \"[2001:0DB8::1]:12345\" | stripDomain }}",
output: "[2001:0DB8::1]:12345",
},
{
// Value can't be split into host and port.
text: "{{ \"[2001:0DB8::1]::12345\" | stripDomain }}",
output: "[2001:0DB8::1]::12345",
},
})
}
type scenario struct {
text string
output string
input interface{}
options []string
queryResult promql.Vector
shouldFail bool
html bool
errorMsg string
}
func testTemplateExpansion(t *testing.T, scenarios []scenario) {
extURL, err := url.Parse("http://testhost:9090/path/prefix")
if err != nil {
panic(err)
}
for _, s := range scenarios {
queryFunc := func(_ context.Context, _ string, _ time.Time) (promql.Vector, error) {
return s.queryResult, nil
}
var result string
var err error
expander := NewTemplateExpander(context.Background(), s.text, "test", s.input, 0, queryFunc, extURL, s.options)
if s.html {
result, err = expander.ExpandHTML(nil)
} else {
result, err = expander.Expand()
}
if s.shouldFail {
require.Error(t, err, "%v", s.text)
require.EqualError(t, err, s.errorMsg)
continue
}
require.NoError(t, err)
if err == nil {
require.Equal(t, s.output, result)
}
}
}
func Test_floatToTime(t *testing.T) {
type args struct {
v float64
}
tests := []struct {
name string
args args
want *time.Time
wantErr bool
}{
{
"happy path",
args{
v: 1657155181,
},
func() *time.Time {
tm := time.Date(2022, 7, 7, 0, 53, 1, 0, time.UTC)
return &tm
}(),
false,
},
{
"more than math.MaxInt64",
args{
v: 1.79769313486231570814527423731704356798070e+300,
},
nil,
true,
},
{
"less than math.MinInt64",
args{
v: -1.79769313486231570814527423731704356798070e+300,
},
nil,
true,
},
}
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
got, err := floatToTime(tt.args.v)
if (err != nil) != tt.wantErr {
t.Errorf("floatToTime() error = %v, wantErr %v", err, tt.wantErr)
return
}
if !reflect.DeepEqual(got, tt.want) {
t.Errorf("floatToTime() got = %v, want %v", got, tt.want)
}
})
}
}