yuanbiao
/
emqx
1
0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
emqx/changes/v4.4.17-en.md

3.0 KiB
Raw Permalink Blame History

v4.4.17

Enhancements

  • Remove the error logs from the listeners with Proxy Protocol enabled when receiving TCP port probes emqx/esockd#172.

    Before this change, if the listener has enabled the Proxy Protocol (listener.tcp.external.proxy_protocol=on), and if the connection get disconnected right after the TCP handshake is completed and before the ProxyInfo is received, the following error log will be printed:

    [error] supervisor: 'esockd_connection_sup - <0.3265.0>', errorContext: connection_shutdown, reason: {recv_proxy_info_error,tcp_closed}, offender:

    After this change, no logs will be printed, but you can still view the statistics of error causes by using the emqx_ctl listeners command.

  • Improve the logs when the listener experiences file descriptor exhaustion errors emqx/esockd#173. Previous log messages:

    [error] Accept error on 0.0.0.0:1883: emfile

    Log messages after this change:

    [error] Accept error on 0.0.0.0:1883: EMFILE (Too many open files)

  • Improve the performance of the rule engine when there are a large number of rules #10283 Before this change, when there were a large number of rules, the execution of the rule engine became a bottleneck, as the rule engine consumed a lot of CPU time on rule queries and matching. In this optimization, we simply added a cache to the rule list, which greatly improved the efficiency of rule execution in this scenario. In our tests, we created 700 rules that did not perform any actions (bound to the "do_nothing" debugging action) on a 32-core, 32GB virtual machine, and sent MQTT messages to EMQX at a rate of 1000 messages per second (i.e. a rule trigger frequency of 700 * 1000 times per second). After applying this optimization, the CPU usage of the rule engine dropped to 55% to 60% of its original level.

Bug fixes

  • Fix that Erlang distribution can't use TLS #9981. About Erlang distribution, See here for details.

  • Fixed MQTT bridge TLS connection could not verify wildcard certificate from peer #10094.

  • Fixed the problem that EMQX could not timely clear the disconnected MQTT connection information due to a large number of retained messages #10189. Before this fix, the emqx_retainer plugin and the connection cleanup process of EMQX shared the same process pool. Therefore, if the process pool was blocked by a large number of retain messages, many disconnected MQTT connections would not be cleaned up in time. For more details, see #9409. In this fix, we created a separate process pool for the emqx_retainer plugin to avoid this problem.

  • Fixed the error in the template file path of Helm Chart.#10229