lafirest
@gbrehmer @wivwiv Count the total of all mqueue lengths in real-time may be very expensive, but we could set a high water marker, and warn when the mqueue length is...
Each session has an mqueue
It looks like marker may not be the good solution for your problem, and we are planning to add some functionality to find which client/session that mqueue is full
@wivwiv Yes, a warning will log whenever a message dropped
@gbrehmer , Hi, There are some solutions under the current EMQX: 1. enlarge the mqueue length 2. adjust mqueue_priorities for important topics both docs can find at here: https://docs.emqx.com/en/enterprise/v4.2/configuration/configuration.html#cluster 3....
@xray-bit A workaround is manually to install in alpine by `apk add tzdata`, or crate a new Dockerfile
> @fridayy I think we can just hash the publisher pid as the start offset, although this isn't the best solution
> hi @lafirest I am not sure if I get it (sorry!): Wouldn't you still get duplicate start offsets (as long as the modulo approach is used)? And messages would...
@IanChiu628 Hi, The user interface and documents of the limiter are still at work. The limiter of the EMQX 5.0 is the C/S architecture. The configuration of the limiter only...
> Hi @lafirest, Thanks for looking into it. I do find the `limiter` field in `listeners.tcp.$name`. It seems like we should put the bucket_name of the limiter into this `limiter`...