hongliang
hongliang
@lsasta 使用异步的方式更多的是用于提高服务端的性能, 同时间接也提升了kafka的写入效率, 如果对数据量很大且性能要求比较高又允许极端异常情况下数据可以有丢失的场景可以使用, 如果不允许数据丢失建议用同步, 另外如果kafka故障建议业务方可以做适当的降级或者重试逻辑来解决。
@agentzh All fixed. And also add some test cases.
@thibaultcha Your comments are very good. But it doesn't apply to us. For multiple apps, which have their only `lua_package_path`, the `lua_package_path` directive and `LUA_PATH` environment variable can be used...
@spacewander Your analysis is very good. This can lead to a conflict among apps. @agentzh Yeah, allowing multiple `init_by_lua*` directive instances is more versatile and flexible, i will work for...
@p0pr0ck5 Hello, will you have a time to re-implement this feature via lua-resty-core's FFI method?
@p0pr0ck5 Hello, Yeah, LGTM, thanks!
@agentzh Thanks for your suggestion, do you have time to help me with this?
@agentzh Okay, rebased :)
@agentzh Can this PR be merged into the master?