elastic/logstash

LogStash should call shutdown hooks

Open

#5,174 建立於 2016年4月22日

在 GitHub 查看
 (2 留言) (0 反應) (0 負責人)Ruby (14,197 star) (3,496 fork)batch import
bughelp wanted

描述

Currently, LogStash will shutdown by waiting for events to flush out and then call do_close on plugins. If --allow-unsafe-shutdown is specified then LogStash shutdown watcher will call Kernel.exit after some time.

However, ZeroMQ plugins will block under certain condition (this is by design of ZeroMQ), thus events will never flush out and do_close will never be called. The way to unblock the sockets is by terminating the ZeroMQ context. This can be done allowing a shutdown hook to be called before exiting forcefully.

貢獻者指南

技術棧
無
領域
backend
議題類型
feature
難度面向新貢獻者的預計實作難度，1 表示很小改動，5 表示專家級工作。
3
預計時間有經驗貢獻者完成調查、實作、測試並準備 pull request 的粗略時間範圍。
1-2 days
活動狀態議題目前的可參與程度：新鮮、活躍、陳舊、阻塞或等待維護者輸入。
stale
清晰度議題是否清楚說明預期改動、驗收標準和下一步。
clear
前置要求
LogStash plugin architectureZeroMQ socket lifecycle
新手友善度1-100 的估計分數，表示該議題對首次貢獻者的友善程度。
30
研究方向
Investigate the LogStash shutdown sequence in the main runner code (e.g., 'logstash/runner.rb' or 'logstash/agent.rb'). Look into how currently 'do close' is called on plugins and where 'Kernel.exit' is invoked. Then design a shutdown hook mechanism that allows ZeroMQ context termination before forceful exit. Check the ZeroMQ Ruby bindings for context termination API.