Justin, That is again a helpful answer, and it works, thx!
The manual says (in https://activemq.apache.org/components/artemis/documentation/latest/large-messages.html): > Warning: setting amqpMinLargeMessageSize to -1, your AMQP message might be > stored as a Core Large Message if the size of the message does not fit into > the journal. > This is the former semantic of the broker and it is kept this way for > compatibility reasons. I ignored this statement because I had just set amqpMinLargeMessageSize to a high value (and not -1). But your answer mentions (almost) the same effect. Shouldn't this warning then be adjusted or extended? e. -----Oorspronkelijk bericht----- Van: Justin Bertram <jbert...@apache.org> Verzonden: donderdag 5 augustus 2021 04:55 Aan: users@activemq.apache.org Onderwerp: Re: large amqp message size for Artemis question EXTERNAL SENDER: Do not click any links or open any attachments unless you trust the sender and know the content is safe. EXPÉDITEUR EXTERNE: Ne cliquez sur aucun lien et n’ouvrez aucune pièce jointe à moins qu’ils ne proviennent d’un expéditeur fiable, ou que vous ayez l'assurance que le contenu provient d'une source sûre. There is a limit where the broker *must* treat any message as a large message and that limit is the journal-buffer-size. If the message can't fit in the journal buffer then it can't be persisted to the journal so it must be stored as a large message. By default the journal-buffer-size is 490 * 1024. Try increasing the journal-buffer-size until you get the behavior you want. Justin On Wed, Aug 4, 2021 at 4:07 PM Dondorp, Erwin <erwin.dond...@cgi.com> wrote: > Hello, > > I'm testing with big amqp messages, which are required to remain > in-memory on the Artemis broker. > The acceptor parameter amqpMinLargeMessageSize is raised from its > default value of 100KiB, to the new value 2000KiB. > Testing is done by just looking at the data/large-messages directory, > which then should not have any files. > > And indeed, messages that have a payload larger than 100KiB now remain > non-large messages. > But when producing larger and larger payloads, message payloads from > 1002153 bytes (I might be one or two bytes off) and bigger still > become large messages. > > But I have no clue on which mechanism causes the decision to make it a > large message as, beyond the acceptor, I did not find any other mechanism. > Normally, the value of the observed boundary gives me a clue, but I > also just do not recognize this number. > > Does anyone have a hint on what to look for? > > thx, > Erwin >