https://bugs.koha-community.org/bugzilla3/show_bug.cgi?id=41506
--- Comment #26 from Tomás Cohen Arazi (tcohen) <[email protected]> --- (In reply to Jonathan Druart from comment #25) > (In reply to Tomás Cohen Arazi (tcohen) from comment #23) > > (In reply to Katrin Fischer from comment #21) > > > (In reply to Mark Hofstetter from comment #20) > > > > will do (as it's our suggestion) > > > > > > Added you to the new bug 42077 as discussed at dev meeting (having it > > > separate from the discussion about removal) > > > > That one is PQA now, and I filed and implemented bug 43729 last night. We > > should probably close this bug report now. > > Bug 43729 is "Remove STOMP/RabbitMQ background jobs support". However I > don't see a clear consensus on this decision in this bug report. I cannot spend time fighting this cause, so this is all I want to say about the matter, and am happy with only getting bug 42077 merged. People have expressed this in different ways and through different channels: virtually nobody seems to be using RabbitMQ. I think we should remove it. If we ever decide to reintroduce a message broker (we should!), we should do so as part of a properly designed architecture, with at least a dedicated task manager service and message dispatching completely decoupled from controller code. That should be a separate effort, undertaken _after_ removing the current implementation. The existing RabbitMQ integration has also been holding back improvements to the background job system. For example, delayed message delivery requires an additional plugin that needs to be manually compiled and installed, making it unsuitable as a core dependency. This limitation was already apparent before we started discussing retries and execution thresholds. Ultimately, we shouldn't let an effectively unused integration dictate how our background job infrastructure evolves. P.S. I think what the project is really missing is a proper task manager. It could initially rely entirely on DB polling, with task dispatching kept separate from task management. That's another reason I believe the current implementation should simply go away: it's too convoluted as it stands, and we would be better off building on a simpler foundation. -- You are receiving this mail because: You are watching all bug changes. _______________________________________________ Koha-bugs mailing list -- [email protected] To unsubscribe send an email to [email protected] website : http://www.koha-community.org/ git : http://git.koha-community.org/ bugs : http://bugs.koha-community.org/
