Heap: A sampling of memory allocations of live objects. You can specify the gc GET parameter to run GC before taking the heap sample.
添加gc这个parameter到get的url里， 可以在采样之前强制执行一次GC后，再进行采样；来之pprof.go(net/http/pprof.go)源码
Heap的手工埋点
pprof.WriteHeapProfile if err := pprof.WriteHeapProfile(f1); err != nil {
panic("could not start heap-1 profile: ")
}
pprof工具Heap分析
运行命令
go tool pprof http://localhost:8999/debug/pprof/heap
使用top命令，查看内存占用最多的地方
Top命令的结果，列表出了在采样的快照，发生的程序中各种函数调用过程中内存占用最多的函数调用；
其中每一行都表示一个采用到的函数调用，
每类都对应着采集到的数据分析； 列的参照如下
flat：函数在调用中内存占用数量
flat%：函数在调用中内存占用的百分比
sum%：所有函数累加使用内存的比例，即所有flat%的总和
cum： 函数以及子函数运行所使用内存，应该大于等于flat
cum%： 函数以及子函数运行所使用内存的比例，应该大于等于flat%
函数的名字
这里可以看到调用现在内存占用最多的占用内存60MB, 占所有内存总量的99.15%，main.startHttpServer.func3， 这个函数的调用占用的内存基本上就是所有的内存， 这个函数究竟做了哪些动作，导致内存的占用， 我们可以在pprof里运行list命令， 进一步进行分析
使用pprof命令：list 函数名。 可以用pprof分析函数中的哪一行导致的内存占用
(pprof) list func3
Total: 60.52MB
ROUTINE ======================== main.startHttpServer.func3 in E:\WORK\PROJECT\git\go\gohutool\boot4go-docker-ui\docker.ui.go
60MB 60MB (flat, cum) 99.15% of Total
. . 180:
. . 181: if len(testPool) >= POOLSIZE {
. . 182: testPool = testPool[0:0]
. . 183: }
. . 184:
60MB 60MB 185: testPool = append(testPool, make([]byte, SIZE))
. . 186:
. . 187: return nil
. . 188: })
. . 189:
. . 190: router.Any("/debug/pprof/*", func(ctx *routing.Context) error {通过这个命令，可以直接追击到源代码的行数，从而进行仔细的调用过程分析
代码185行: make([]byte, SIZE) 进行了内存的操作， 这里的60M全部这里进行操作的。
使用图形化web命令进行分析
在pprof的交互模式里输入web命令
使用tree命令查看
(pprof) tree
Showing nodes accounting for 206.50MB, 99.28% of 208MB total
Dropped 15 nodes (cum <= 1.04MB)
----------------------------------------------------------+-------------
flat flat% sum% cum cum% calls calls% + context
----------------------------------------------------------+-------------
204MB 100% | github.com/qiangxue/fasthttp-routing.(*Context).Next
204MB 98.08% 98.08% 204MB 98.08% | main.startHttpServer.func3
----------------------------------------------------------+-------------
2.50MB 100% | runtime.newm
2.50MB 1.20% 99.28% 2.50MB 1.20% | runtime.allocm
----------------------------------------------------------+-------------
204MB 100% | github.com/qiangxue/fasthttp-routing.(*Router).HandleRequest
0 0% 99.28% 204MB 98.08% | github.com/qiangxue/fasthttp-routing.(*Context).Next
204MB 100% | main.startHttpServer.func3
----------------------------------------------------------+-------------
204MB 100% | main.startHttpServer.func8
0 0% 99.28% 204MB 98.08% | github.com/qiangxue/fasthttp-routing.(*Router).HandleRequest
204MB 100% | github.com/qiangxue/fasthttp-routing.(*Context).Next
----------------------------------------------------------+-------------
204MB 100% | github.com/valyala/fasthttp.(*workerPool).workerFunc
0 0% 99.28% 204MB 98.08% | github.com/valyala/fasthttp.(*Server).serveConn
204MB 100% | main.startHttpServer.func8
----------------------------------------------------------+-------------
0 0% 99.28% 204MB 98.08% | github.com/valyala/fasthttp.(*workerPool).getCh.func1
204MB 100% | github.com/valyala/fasthttp.(*workerPool).workerFunc
----------------------------------------------------------+-------------
204MB 100% | github.com/valyala/fasthttp.(*workerPool).getCh.func1
0 0% 99.28% 204MB 98.08% | github.com/valyala/fasthttp.(*workerPool).workerFunc
204MB 100% | github.com/valyala/fasthttp.(*Server).serveConn
----------------------------------------------------------+-------------
204MB 100% | github.com/valyala/fasthttp.(*Server).serveConn
0 0% 99.28% 204MB 98.08% | main.startHttpServer.func8
204MB 100% | github.com/qiangxue/fasthttp-routing.(*Router).HandleRequest
----------------------------------------------------------+-------------
0 0% 99.28% 2MB 0.96% | runtime.mstart
2MB 100% | runtime.mstart0
----------------------------------------------------------+-------------
2MB 100% | runtime.mstart
0 0% 99.28% 2MB 0.96% | runtime.mstart0
2MB 100% | runtime.mstart1
----------------------------------------------------------+-------------
2MB 100% | runtime.mstart0
0 0% 99.28% 2MB 0.96% | runtime.mstart1
2MB 100% | runtime.schedule
----------------------------------------------------------+-------------
2.50MB 100% | runtime.startm
0 0% 99.28% 2.50MB 1.20% | runtime.newm
2.50MB 100% | runtime.allocm
----------------------------------------------------------+-------------
2.50MB 100% | runtime.schedule
0 0% 99.28% 2.50MB 1.20% | runtime.resetspinning
2.50MB 100% | runtime.wakep
----------------------------------------------------------+-------------
2MB 80.00% | runtime.mstart1
0 0% 99.28% 2.50MB 1.20% | runtime.schedule
2.50MB 100% | runtime.resetspinning
----------------------------------------------------------+-------------
2.50MB 100% | runtime.wakep
0 0% 99.28% 2.50MB 1.20% | runtime.startm
2.50MB 100% | runtime.newm
----------------------------------------------------------+-------------
2.50MB 100% | runtime.resetspinning
0 0% 99.28% 2.50MB 1.20% | runtime.wakep
2.50MB 100% | runtime.startm
----------------------------------------------------------+-------------对比一下func3和func8
(pprof) list func3
Total: 208MB
ROUTINE ======================== main.startHttpServer.func3 in E:\WORK\PROJECT\git\go\gohutool\boot4go-docker-ui\docker.ui.go
204MB 204MB (flat, cum) 98.08% of Total
. . 180:
. . 181: if len(testPool) >= POOLSIZE {
. . 182: testPool = testPool[0:0]
. . 183: }
. . 184:
204MB 204MB 185: testPool = append(testPool, make([]byte, SIZE))
. . 186:
. . 187: return nil
. . 188: })
. . 189:
. . 190: router.Any("/debug/pprof/*", func(ctx *routing.Context) error {(pprof) list func8
Total: 208MB
ROUTINE ======================== main.startHttpServer.func8 in E:\WORK\PROJECT\git\go\gohutool\boot4go-docker-ui\docker.ui.go
0 204MB (flat, cum) 98.08% of Total
. . 248:
. . 249: Logger.Debug("router.HandleRequest is finish")
. . 250:
. . 251: }()
. . 252:
. 204MB 253: router.HandleRequest(ctx)
. . 254: }
. . 255:
. . 256: //go http.ListenAndServe("0.0.0.0:8887", nil)
. . 257:
. . 258: // Start HTTP server.
heap提供了对内存进行分析的途径，通过对内存的占用量的分析，我们可以结合着程序的压力测试，对程序的内存泄漏进行排查和分析，对内存只增不减的等可疑情况地方，进行排查，最终发现问题
heap的pprof样本中的数据项
技巧这些信息是样本中的数据项，相对于我们通过web访问时，访问当前时间的heap的一个快照； 熟悉这些数据项所表示的含义，也非常对我们了解当前程序运行时的内存情况非常有帮助。
net/http/pprof/pprof.goruntime/pprof/pprof.goruntime/mstats.go
