From e364249ae9e7b8fad751d53308e461c183a938e3 Mon Sep 17 00:00:00 2001 From: Meggielqk <126552073+Meggielqk@users.noreply.github.com> Date: Fri, 8 Nov 2024 10:57:29 +0800 Subject: [PATCH] Update en_US/changes/changes-ee-v5.md Co-authored-by: zmstone --- en_US/changes/changes-ee-v5.md | 13 ------------- 1 file changed, 13 deletions(-) diff --git a/en_US/changes/changes-ee-v5.md b/en_US/changes/changes-ee-v5.md index 4249862d8..aba2921c9 100644 --- a/en_US/changes/changes-ee-v5.md +++ b/en_US/changes/changes-ee-v5.md @@ -14,19 +14,6 @@ Make sure to check the breaking changes and known issues before upgrading to EMQ **Note:** This change does not apply to RHEL 7/CentOS 7, as they continue to use OpenSSL 1.0.x. -- [#14047](https://github.com/emqx/emqx/pull/14047) Lowered default `active_n` value from `100` to `10`. - - This change improves the responsiveness of MQTT clients to control signals, particularly when publishing at high rates with small messages. - - The new `active_n` value of `10` is set deliberately lower than the default Receive-Maximum (`32`), to introduce more push-back at the TCP layer in the following scenarios: - - - The MQTT client process is blocked while performing external authorization checks. - - The MQTT client process is blocked during data integration message sends. - - EMQX is experiencing overload conditions. - - Performance testing showed no significant increase in latency across various scenarios (one-to-one, fan-in, and fan-out) on 8-core, 16GB memory nodes. - However, on 2-core, 4GB memory nodes, the baseline latency (with active_n = `100`) was already in the higher 3-digit range with high CPU utilization. - The decision to lower `active_n` optimizes for more common use cases where system stablity takes precedence over latency (in smaller instances). - [#14059](https://github.com/emqx/emqx/pull/14059) Added a new configuration option for the retainer to cap message expiry intervals for retained messages. This enables garbage collection to remove messages sooner if storage is running low.