ethfoo
find another: ``` fatal error: concurrent map iteration and map write goroutine 346 [running]: runtime.throw(0x21ac606, 0x26) /usr/local/go/src/runtime/panic.go:1117 +0x72 fp=0xc000a76e48 sp=0xc000a76e18 pc=0x440b52 runtime.mapiternext(0xc000b069c0) /usr/local/go/src/runtime/map.go:858 +0x54c fp=0xc000a76ec8 sp=0xc000a76e48 pc=0x4196cc reflect.mapiternext(0xc000b069c0) /usr/local/go/src/runtime/map.go:1346 +0x2b...
Is it possible to replace this zinc sink with an HTTP sink ?
> 已按建议修改了所有的代码 @ethfoo 又加了几个review comment，麻烦再看下
@shaoyun 是否在微信群里？如果不在的话麻烦加一下：https://loggie-io.github.io/docs/getting-started/overview/#_3
> How about just use the slice of k/v pair instead of map? In some interceptors, we need to process or transform specific key-value, map is more suitable than the...
> Yes. But it would become the bottleneck of one single pipeline. I had no idea which sink you'd put in a pipeline, but the sink might be the bottleneck...
1. isdrop感觉可以换个名字，不是特别直观 2. i.isDrop不需要传进方法里，可以直接从interceptor config里拿
麻烦发一下你的配置，和捞取一些采集的日志看下
@lyp256 来吧
protobuf解码需要引入固定类型的proto文件吧？