Commits on Dec 9, 2020

1. mqtt_proxy: Execute retry_connect in main thread …

   timer_execute will execute registered method in main thread.
   Prevous implementation will causes unhandled retrying with the following error sequence:
   
   1. send_packet(MQTT::Packet::Subscribe) in mqtt_proxy.rb (running on `@connection_thread`)
   2. When https://github.com/njh/ruby-mqtt/blob/v0.5.0/lib/mqtt/client.rb#L554, and then keep_alive! method failed in client.rb (running on `@read_thread`)
   3. IOError occurred in https://github.com/njh/ruby-mqtt/blob/v0.5.0/lib/mqtt/client.rb#L556 in mqtt_proxy.rb (running on `@connection_thread`)
   4. Executed #retry_connect in mqtt_proxy.rb (running on `@connection_thread`)
   5. Before reached in https://github.com/toyokazu/fluent-plugin-mqtt-io/blob/master/lib/fluent/plugin/mqtt_proxy.rb#L93, `@retry_interval` second(s) passes in mqtt_proxy.rb (running on `@connection_thread`)
   6. During 5. execution, https://github.com/toyokazu/fluent-plugin-mqtt-io/blob/master/lib/fluent/plugin/mqtt_proxy.rb#L152 is executed in main thread
   7. Then, retry is disabled.
   
   In short, `@connection_thread` in mqtt_proxy.rb and main thread will
   causes updating `@_retrying` race condition.
   
   To prevent this race condition, we should use `timer_execute` for
   `#retry_connect` to execute retrying connection operation on main thread
   not executing it on `@connection_thread`.
   This race condition will not happen when every running retrying operation
   runs on main thread.
   
   Signed-off-by: Hiroshi Hatake <hatake@clear-code.com>

   

   cosmo0920 committed Dec 9, 2020
   Configuration menu

   • View commit details
   • Copy full SHA for 3ecae4e
   • Browse repository at this point

   Copy the full SHA

   3ecae4e View commit details

   Browse the repository at this point in the history