[zeromq-dev] Limited UDP Message Size

Stephan Opfer opfer at vs.uni-kassel.de
Thu Jan 16 10:05:35 CET 2020


I would like to, but I need to finish my PhD within ~2 month. So I am 
open for anyone to jump in. ;-)

Otherwise, it will take some month until I can give it a try.

On 14.01.20 13:56, Doron Somech wrote:
> Make sense, do you want to try and make a pull request?
>
> On Tue, Jan 14, 2020 at 11:40 AM Stephan Opfer <opfer at vs.uni-kassel.de 
> <mailto:opfer at vs.uni-kassel.de>> wrote:
>
>     Okay, so then my suggestion would be ~2^16 bytes as maximum value,
>     but a default value of 8191 bytes or lower. Plus the option to
>     change the value, in order to make libzmq applicable in local
>     network applications that use larger UDP messages (e.g. a team of
>     service robots cooperating in your household). Further an API
>     level error message or assert, that shortly explains the issue
>     when sending UDP messages, larger than the value set.
>
>     Thoughts?
>
>     On 12.01.20 23:04, Greg Young wrote:
>>     If not clear ... if you are on a network *you control* is very
>>     different than being on one you do not.
>>
>>     On Sun, Jan 12, 2020 at 5:02 PM Greg Young
>>     <gregoryyoung1 at gmail.com <mailto:gregoryyoung1 at gmail.com>> wrote:
>>
>>         This is correct ... Some things may not accept/forward it.
>>
>>         Beyond that its common for things to get *added to it* thus
>>         you want to be below. Consider a case where its being wrapped
>>         (tunneling is a good example).
>>
>>         Over the internet the maximum completely safe size is even
>>         smaller around 500-600 bytes :O You can usually get above
>>         this but +- 1k is about the limit.
>>
>>
>>         On Sun, Jan 12, 2020 at 4:42 PM Stephan Opfer
>>         <opfer at vs.uni-kassel.de <mailto:opfer at vs.uni-kassel.de>> wrote:
>>
>>             Ah okay. I read some posts about MTU and fragmentation.
>>             So in theory  2^16= 65536 Bytes should be possible, but
>>             not all nodes will accept such large UDP packets and
>>             support the fragmentation of it?
>>
>>             FYI: My use case is that of a local network (IOT +
>>             Autonomous Robots) and increasing the size would help me. :)
>>
>>             On 12.01.20 13:25, Doron Somech wrote:
>>>             We can increase it to around 65,000 bytes, however,
>>>             large UDP packets would probably be dropped when
>>>             traveling outside your local area network.
>>>             The safe size for the UDP packet to travel outside of
>>>             the local area network is around 500 bytes.
>>>             If you do want to use large UDP packets on your local
>>>             area network you might want to increase the network MTU
>>>             (jumbo frames) for better delivery.
>>>
>>>             For example, some AWS instances support 9001 bytes MTU
>>>             for inter-instance communication.
>>>
>>>
>>>             On Fri, Jan 10, 2020 at 3:25 PM Stephan Opfer
>>>             <opfer at vs.uni-kassel.de <mailto:opfer at vs.uni-kassel.de>>
>>>             wrote:
>>>
>>>                 Hi all,
>>>
>>>                 since 2016 there is this issue
>>>                 (https://github.com/zeromq/libzmq/issues/2009) about
>>>                 some limitations of
>>>                 the udp_engine in libzmq.
>>>
>>>                 Is it possible that someone at least will increase
>>>                 the "#define
>>>                 MAX_UDP_MSG 8192" in libzmq/src/udp_engine.hpp to
>>>                 some reasonable value?
>>>                 For example "ZMQ_MAX_MSGSZ" from the zmq context
>>>                 option, which is MAX_INT.
>>>
>>>                 Maybe, someone could spare some time at the
>>>                 Hackathon... ;-)
>>>
>>>                 Greetings,
>>>
>>>                    Stephan
>>>
>>>                 -- 
>>>                 Distributed Systems Research Group
>>>                 Stephan Opfer  T. +49 561 804-6279  F. +49 561 804-6277
>>>                 Univ. Kassel,  FB 16,  Wilhelmshöher Allee 73, 
>>>                 D-34121 Kassel
>>>                 WWW: http://www.uni-kassel.de/go/vs_stephan-opfer/
>>>
>>>                 _______________________________________________
>>>                 zeromq-dev mailing list
>>>                 zeromq-dev at lists.zeromq.org
>>>                 <mailto:zeromq-dev at lists.zeromq.org>
>>>                 https://lists.zeromq.org/mailman/listinfo/zeromq-dev
>>>
>>>
>>>             _______________________________________________
>>>             zeromq-dev mailing list
>>>             zeromq-dev at lists.zeromq.org  <mailto:zeromq-dev at lists.zeromq.org>
>>>             https://lists.zeromq.org/mailman/listinfo/zeromq-dev
>>
>>             -- 
>>             Distributed Systems Research Group
>>             Stephan Opfer  T. +49 561 804-6279  F. +49 561 804-6277
>>             Univ. Kassel,  FB 16,  Wilhelmshöher Allee 73,  D-34121 Kassel
>>             WWW:http://www.uni-kassel.de/go/vs_stephan-opfer/
>>
>>             _______________________________________________
>>             zeromq-dev mailing list
>>             zeromq-dev at lists.zeromq.org
>>             <mailto:zeromq-dev at lists.zeromq.org>
>>             https://lists.zeromq.org/mailman/listinfo/zeromq-dev
>>
>>
>>
>>         -- 
>>         Studying for the Turing test
>>
>>
>>
>>     -- 
>>     Studying for the Turing test
>>
>>     _______________________________________________
>>     zeromq-dev mailing list
>>     zeromq-dev at lists.zeromq.org  <mailto:zeromq-dev at lists.zeromq.org>
>>     https://lists.zeromq.org/mailman/listinfo/zeromq-dev
>
>     -- 
>     Distributed Systems Research Group
>     Stephan Opfer  T. +49 561 804-6279  F. +49 561 804-6277
>     Univ. Kassel,  FB 16,  Wilhelmshöher Allee 73,  D-34121 Kassel
>     WWW:http://www.uni-kassel.de/go/vs_stephan-opfer/
>
>     _______________________________________________
>     zeromq-dev mailing list
>     zeromq-dev at lists.zeromq.org <mailto:zeromq-dev at lists.zeromq.org>
>     https://lists.zeromq.org/mailman/listinfo/zeromq-dev
>
>
> _______________________________________________
> zeromq-dev mailing list
> zeromq-dev at lists.zeromq.org
> https://lists.zeromq.org/mailman/listinfo/zeromq-dev

-- 
Distributed Systems Research Group
Stephan Opfer  T. +49 561 804-6279  F. +49 561 804-6277
Univ. Kassel,  FB 16,  Wilhelmshöher Allee 73,  D-34121 Kassel
WWW: http://www.uni-kassel.de/go/vs_stephan-opfer/

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.zeromq.org/pipermail/zeromq-dev/attachments/20200116/06f8b709/attachment.htm>


More information about the zeromq-dev mailing list