Slide 1
Slide 1 text
Go pro ling
Steve Carrupt
Nulab Software Developer
Slide 2
Slide 2 text
Agenda
pprof
go-torch
Demo
Questions
Slide 3
Slide 3 text
pprof
CPU, heap, blocks, routines pro lers provided by the Go runtime
Get pro ling data through
Benchmark tests
$ go test -v -run=^$ -bench=^ -cpuprofile=prof.cpu
$ go tool pprof [binary].test prof.cpu
pprof HTTP handlers
import _ net/http/pprof
$ go tool pprof http://localhost:8080/debug/pprof/profile
code
runtime.StartCPUPro le or runtime.WriteHeapPro le
Slide 4
Slide 4 text
CLI pro ling example
Slide 5
Slide 5 text
Web pro ling example
Slide 6
Slide 6 text
go-torch
Easier to read pro ling data with ame graphes
Slide 7
Slide 7 text
Demo
github.com/scarrupt/go-pro ling (https://github.com/scarrupt/go-pro ling)
Slide 8
Slide 8 text
Questions
The real problem is that programmers have spent far too much time worrying about
e ciency in the wrong places and at the wrong times; premature optimization is the root of
all evil (or at least most of it) in programming.
Donald Ervin Knuth
sources:
golang.org/pkg/net/http/pprof/ (https://golang.org/pkg/net/http/pprof/)
github.com/uber/go-torch (https://github.com/uber/go-torch)
Slide 9
Slide 9 text
Thank you
Steve Carrupt
Nulab Software Developer
@stevecarrupt (http://twitter.com/stevecarrupt)