Thank you! On Sunday, September 17, 2017 at 11:32:17 PM UTC-5, Michael Klishin wrote: > > The point of lazy queues is not that nothing ever is stored in memory. > It's that queues try to move messages > to disk very aggressively (default mode keeps a portion in RAM using > metrics such as ingress/egress rates and various > configurable values). > > On Sun, Sep 17, 2017 at 10:29 PM, Michael Klishin <mkli...@pivotal.io > <javascript:>> wrote: > >> This is a question for rabbitmq-users but I'm happy to answer it here. >> >> Lazy queues per se don't have a limit but RabbitMQ message store still >> does: when messages >> are sent to the message store [as opposed to being embedded into queue >> index], message store index(es) >> come into play and by default the index is in-memory. >> >> There is a plugin [1] that lets you use LevelDB for the index. Even >> though LevelDB is an efficient >> data store, expect a write throughput hit compared to the default >> implementation. >> >> 1. https://github.com/rabbitmq/rabbitmq-msg-store-index-eleveldb >> >> On Sun, Sep 17, 2017 at 8:25 PM, Steve Suehs <skelt...@gmail.com >> <javascript:>> wrote: >> >>> I've been experimenting with Clojure and RabbitMQ using the langohr >>> library. Thank you for building and sharing it. >>> >>> I've been learning more about the care and feeding of a rabbitmq >>> server. I crashed mine installed on my dev laptop several times by filling >>> a queue with messages. If queues fill up and memory gets tight, a server >>> can be overwhelmed and fall over. If I set or arrange for a reasonable >>> memory limit, things go ok. I am now running RabbitMQ in a Docker container >>> with a memory limit. >>> >>> My latest experiments were with lazy queues, which are supposed to write >>> events to disk and not exhaust memory. I still seem to hit..well, not hit, >>> rather, an asymptotic limit at around 261 million events in the queue. Is >>> this considered outrageously and unreasonably large? Publication rate >>> slowed to 700/s. >>> >>> I also notice that if the consumers are caught up with the producer >>> everything seems to scream at twice the rate than when storage is used. If >>> storage is used I'm seeing about 6000r/s. That's cool...just not exactly >>> what I expected from the description of a lazy queue. >>> >>> >>> <https://lh3.googleusercontent.com/-XSzuKvXxMxI/Wb8uHb7LKmI/AAAAAAAAEYA/G6V5v93_yGcJW6jFPCnQUUBOp1LGCm7nQCLcBGAs/s1600/rabbitmqLimit.png> >>> >>> >>> >>> -- >>> You received this message because you are subscribed to the Google >>> Groups "clojure-rabbitmq" group. >>> To unsubscribe from this group and stop receiving emails from it, send >>> an email to clojure-rabbit...@googlegroups.com <javascript:>. >>> For more options, visit https://groups.google.com/d/optout. >>> >> >> >> >> -- >> MK >> >> Staff Software Engineer, Pivotal/RabbitMQ >> > > > > -- > MK > > Staff Software Engineer, Pivotal/RabbitMQ >
-- You received this message because you are subscribed to the Google Groups "clojure-rabbitmq" group. To unsubscribe from this group and stop receiving emails from it, send an email to clojure-rabbitmq+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.