Enhancements

• #9802 Support HAProxy protocol for HTTP API.

• #9871 Allow the placeholder to be anywhere in the topic for authz rules. e.g: {allow, {username, "who"}, publish, ["t/foo${username}boo/${clientid}xxx"]}.

• #9910 Add start operation to bridges API to allow manual reconnect after failure.

• #9917 Stop building -alpine docker image because it's size is larger than the regular one based on debian slim

• #9930 Expose the stats live_connections.count and live_connections.max to Prometheus.

• #9936 Disable disksup (part of os_mon) in releases by default, no warnings are issued when a disk error occurs.

• #9954 Improve bridge performance

Bug fixes

• #9864 Fix the exclusive topics aren't removed when the session has already been cleaned.

• #9875 Return 400 if a broken plugin package is uploaded from HTTP API, also cleanup if plugin is not accepted.

• #9916 Fix MQTT bridge fails to verify TLS wildcard server certificate.

• #9922 Fix the issue with the bridge resource buffer where it might become stuck if enough async queries fill the inflight window full before failing with retryable errors.

• #9923 Fix REPORT_CB/2 CRASH error logs when errors happen during boot-up or shutdown.

• #9938 Report some egress MQTT bridge errors as recoverable, and thus retryable.

• #9946 Add back reconnect_interval as deprecated field for MQTT bridge. The field was removed from v5.0.16/e5.0.0 by mistake, caused new version unable to start on old config. Now it's added back as deprecated (config value is ignored if provided).

• #9951 Propagate errors from operations (start|stop|restart) on bridges API if called for all nodes.

• #9952 Disallow subscribing with QoS 2 for ingress MQTT bridges. Allow user to configure clean_start option for ingress MQTT bridges, however.