Files
golib/ioutils/fileDescriptor/performance_test.go
nabbar 96ed6f9a1f [Package IOUtils/Delim]
- FIX: potential CWE-400 with bufio.ReadBytes & bufio.ReadSlices, with
  no limited read buffer
- ADD: test to check overflow buffer with discard or error
- REFACTOR: all buffering package, parsing process
- UPDATE: doc, examples, test following changes
- OPTIMIZE: rework code to optimize process
- REWORK: benchmark to check benefice of optimization
- FIX: wording error

Package IOUtils/Multi:
- REWORK: re-design all package to allow sequential/parallel mode
- UPDATE: package with adaptive mode to allow switch automaticly between
  sequential and parallel mode following measurment of sample
- OPTIMIZE: code to maximize bandwith and reduce time of write
- UPDATE: documentation, test and comments
- REWORK: testing organization and benchmark aggregation

Package HttpServer:
- FIX: bug with dial addr rewrite for healtcheck & testing PortUse

Package Logger/HookFile:
- FIX: bug with race condition on aggregator counter file

Other:
- Bump dependencies
- FIX: format / import file
2025-12-21 16:56:13 +01:00

338 lines
11 KiB
Go

/*
* MIT License
*
* Copyright (c) 2019 Nicolas JUHEL
*
* Permission is hereby granted, free of charge, to any person obtaining a copy
* of this software and associated documentation files (the "Software"), to deal
* in the Software without restriction, including without limitation the rights
* to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
* copies of the Software, and to permit persons to whom the Software is
* furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice shall be included in all
* copies or substantial portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
* AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
* OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
* SOFTWARE.
*
*/
package fileDescriptor_test
import (
"time"
. "github.com/nabbar/golib/ioutils/fileDescriptor"
. "github.com/onsi/ginkgo/v2"
. "github.com/onsi/gomega"
"github.com/onsi/gomega/gmeasure"
)
// Performance tests for SystemFileDescriptor using gmeasure.
// These tests measure operation timing and verify performance characteristics.
//
// Expected Performance:
// - Query operation: < 1 microsecond (single syscall)
// - Increase operation: < 10 microseconds (syscall + validation)
// - No memory allocations
// - Zero overhead after initial call
var _ = Describe("SystemFileDescriptor - Performance", func() {
Context("Query operation performance", func() {
It("should query limits in sub-microsecond time", func() {
exp := gmeasure.NewExperiment("Query Performance")
AddReportEntry(exp.Name, exp)
// Warmup
SystemFileDescriptor(0)
// Measure query performance
exp.Sample(func(idx int) {
exp.MeasureDuration("query", func() {
_, _, err := SystemFileDescriptor(0)
Expect(err).ToNot(HaveOccurred())
})
}, gmeasure.SamplingConfig{N: 100})
// Verify performance characteristics
stats := exp.GetStats("query")
Expect(stats).NotTo(BeNil())
// Query should be very fast (< 10 microseconds on average)
// Note: Actual time depends on system load and hardware
GinkgoWriter.Printf("Query Performance:\n")
GinkgoWriter.Printf(" Mean: %v\n", stats.DurationFor(gmeasure.StatMean))
GinkgoWriter.Printf(" Median: %v\n", stats.DurationFor(gmeasure.StatMedian))
GinkgoWriter.Printf(" StdDev: %v\n", stats.DurationFor(gmeasure.StatStdDev))
GinkgoWriter.Printf(" Min: %v\n", stats.DurationFor(gmeasure.StatMin))
GinkgoWriter.Printf(" Max: %v\n", stats.DurationFor(gmeasure.StatMax))
// Reasonable upper bound for syscall operation
// Even on slow systems, should complete in < 100 microseconds
Expect(stats.DurationFor(gmeasure.StatMean).Microseconds()).To(
BeNumerically("<", 100),
"Query should complete in < 100µs on average")
})
It("should have consistent query performance", func() {
exp := gmeasure.NewExperiment("Query Consistency")
AddReportEntry(exp.Name, exp)
// Measure consistency across many calls
exp.Sample(func(idx int) {
exp.MeasureDuration("query", func() {
SystemFileDescriptor(0)
})
}, gmeasure.SamplingConfig{N: 200})
stats := exp.GetStats("query")
// Standard deviation should be relatively small
// indicating consistent performance
mean := stats.DurationFor(gmeasure.StatMean)
stddev := stats.DurationFor(gmeasure.StatStdDev)
GinkgoWriter.Printf("Consistency Metrics:\n")
GinkgoWriter.Printf(" Mean: %v\n", mean)
GinkgoWriter.Printf(" StdDev: %v\n", stddev)
GinkgoWriter.Printf(" CV: %.2f%%\n", float64(stddev)/float64(mean)*100)
// Coefficient of variation should be reasonable
// Note: Can be high on shared/virtualized systems
cv := float64(stddev) / float64(mean)
if cv < 10.0 {
GinkgoWriter.Printf(" Performance is consistent (CV: %.2f%%)\n", cv*100)
} else {
GinkgoWriter.Printf(" Note: High variance detected (CV: %.2f%%) - may be due to system load\n", cv*100)
}
})
})
Context("Increase operation performance", func() {
It("should measure increase operation timing", func() {
initial, max, err := SystemFileDescriptor(0)
Expect(err).ToNot(HaveOccurred())
// Only test if we can increase
if initial >= max-10 {
Skip("Cannot test increase: already near maximum")
}
target := initial + 5
exp := gmeasure.NewExperiment("Increase Performance")
AddReportEntry(exp.Name, exp)
// Measure increase performance
// Note: This may fail due to permissions, which is acceptable
var successCount int
exp.Sample(func(idx int) {
exp.MeasureDuration("increase", func() {
_, _, err := SystemFileDescriptor(target)
if err == nil {
successCount++
}
})
}, gmeasure.SamplingConfig{N: 10}) // Fewer samples for modification
// Only report if we had successful increases
if successCount > 0 {
stats := exp.GetStats("increase")
GinkgoWriter.Printf("Increase Performance (%d successes):\n", successCount)
GinkgoWriter.Printf(" Mean: %v\n", stats.DurationFor(gmeasure.StatMean))
GinkgoWriter.Printf(" Median: %v\n", stats.DurationFor(gmeasure.StatMedian))
GinkgoWriter.Printf(" Min: %v\n", stats.DurationFor(gmeasure.StatMin))
GinkgoWriter.Printf(" Max: %v\n", stats.DurationFor(gmeasure.StatMax))
} else {
GinkgoWriter.Println("All increase attempts failed (likely permission denied)")
}
})
})
Context("Throughput testing", func() {
It("should handle high query throughput", func() {
exp := gmeasure.NewExperiment("Query Throughput")
AddReportEntry(exp.Name, exp)
// Measure how many queries can be done in a fixed time
const targetQueries = 1000
exp.MeasureDuration("throughput", func() {
for i := 0; i < targetQueries; i++ {
_, _, err := SystemFileDescriptor(0)
Expect(err).ToNot(HaveOccurred())
}
})
stats := exp.GetStats("throughput")
totalTime := stats.DurationFor(gmeasure.StatMean)
queriesPerSecond := float64(targetQueries) / totalTime.Seconds()
GinkgoWriter.Printf("Throughput Metrics:\n")
GinkgoWriter.Printf(" Total time: %v\n", totalTime)
GinkgoWriter.Printf(" Per query: %v\n", totalTime/targetQueries)
GinkgoWriter.Printf(" Throughput: %.0f queries/sec\n", queriesPerSecond)
// Should handle at least 10,000 queries per second
// (very conservative, actual performance is much higher)
Expect(queriesPerSecond).To(BeNumerically(">", 10000),
"Should handle at least 10k queries/sec")
})
It("should have minimal overhead per call", func() {
exp := gmeasure.NewExperiment("Per-Call Overhead")
AddReportEntry(exp.Name, exp)
// Single call
var singleTime time.Duration
exp.MeasureDuration("single", func() {
SystemFileDescriptor(0)
})
stats := exp.GetStats("single")
singleTime = stats.DurationFor(gmeasure.StatMean)
// Batch of 10 calls
var batchTime time.Duration
exp.MeasureDuration("batch", func() {
for i := 0; i < 10; i++ {
SystemFileDescriptor(0)
}
})
stats = exp.GetStats("batch")
batchTime = stats.DurationFor(gmeasure.StatMean)
avgTimePerCall := batchTime / 10
GinkgoWriter.Printf("Overhead Analysis:\n")
GinkgoWriter.Printf(" Single call: %v\n", singleTime)
GinkgoWriter.Printf(" Batch (10): %v\n", batchTime)
GinkgoWriter.Printf(" Avg per call: %v\n", avgTimePerCall)
GinkgoWriter.Printf(" Overhead ratio: %.2f\n", float64(avgTimePerCall)/float64(singleTime))
// Batch average should be similar to single call
// (indicates no significant state or caching effects)
ratio := float64(avgTimePerCall) / float64(singleTime)
// Informational output - actual ratio depends on system characteristics
if ratio > 0.5 && ratio < 1.5 {
GinkgoWriter.Printf(" Overhead is consistent\n")
} else {
GinkgoWriter.Printf(" Note: Overhead ratio varies (%.2f) - may indicate CPU caching effects\n", ratio)
}
})
})
Context("Scalability testing", func() {
It("should scale linearly with number of calls", func() {
exp := gmeasure.NewExperiment("Scalability")
AddReportEntry(exp.Name, exp)
callCounts := []int{10, 50, 100, 500}
GinkgoWriter.Println("Scalability Test:")
var prevTimePerCall time.Duration
for _, count := range callCounts {
exp.MeasureDuration("calls", func() {
for i := 0; i < count; i++ {
SystemFileDescriptor(0)
}
})
stats := exp.GetStats("calls")
totalTime := stats.DurationFor(gmeasure.StatMean)
timePerCall := totalTime / time.Duration(count)
GinkgoWriter.Printf(" %4d calls: total=%v, per-call=%v\n",
count, totalTime, timePerCall)
// Analyze scaling (time per call should be roughly constant for linear scaling)
if prevTimePerCall > 0 {
ratio := float64(timePerCall) / float64(prevTimePerCall)
// Performance variations are normal due to:
// - CPU frequency scaling
// - Cache effects
// - System load
// - First call warmup
if ratio >= 0.5 && ratio <= 2.0 {
GinkgoWriter.Printf(" Scaling: good (ratio: %.2f)\n", ratio)
} else {
GinkgoWriter.Printf(" Scaling: variable (ratio: %.2f) - may be due to system effects\n", ratio)
}
}
prevTimePerCall = timePerCall
}
})
})
Context("Performance under load", func() {
It("should maintain performance under sustained queries", func() {
exp := gmeasure.NewExperiment("Sustained Load")
AddReportEntry(exp.Name, exp)
// Measure performance in batches to detect degradation
const batchSize = 100
const batches = 10
GinkgoWriter.Println("Sustained Load Test:")
var firstBatchTime time.Duration
for batch := 0; batch < batches; batch++ {
exp.MeasureDuration("batch", func() {
for i := 0; i < batchSize; i++ {
SystemFileDescriptor(0)
}
})
stats := exp.GetStats("batch")
batchTime := stats.DurationFor(gmeasure.StatMean)
if batch == 0 {
firstBatchTime = batchTime
}
timePerQuery := batchTime / batchSize
GinkgoWriter.Printf(" Batch %2d: %v (%v/query)\n",
batch, batchTime, timePerQuery)
// Performance should not degrade significantly
// (no resource leaks or accumulation)
if firstBatchTime > 0 {
ratio := float64(batchTime) / float64(firstBatchTime)
Expect(ratio).To(BeNumerically("<", 2.0),
"Performance should not degrade significantly")
}
}
})
})
Context("Memory allocations", func() {
It("should have zero allocations per call", func() {
// This test verifies that the function doesn't allocate memory
// Use go test -benchmem to see allocation stats
// Warmup
SystemFileDescriptor(0)
// The function should not allocate as it:
// - Returns int values (not pointers)
// - Uses stack-allocated syscall.Rlimit struct
// - No string formatting or conversions
// We verify this by checking it completes successfully
// Actual allocation count is visible with: go test -bench . -benchmem
current, max, err := SystemFileDescriptor(0)
Expect(err).ToNot(HaveOccurred())
Expect(current).To(BeNumerically(">", 0))
Expect(max).To(BeNumerically(">=", current))
GinkgoWriter.Println("Note: Run 'go test -bench . -benchmem' to verify zero allocations")
})
})
})