You nailed it. There’s also the issue of accounting. To basically find out for a given IP how backlogged is it…
Also, I pushed the new version of our scheduler that uses the ability to release queues. Working great so far. We went from an unbounded, ever growing list of queues, some up to 150k or more.. down to about 3000 at peak. They’re getting GCd every minute and for the most part not locking up. I’m sure I have a few straggling bugs but ActiveMQ for the most part is handling it. Knock on wood :) On Fri, Feb 6, 2015 at 2:30 PM, artnaseef <[email protected]> wrote: > I think I understand - there's a need to throttle for each IP, and there > are > many IP addresses, and the set of IP addresses changes over time. > > It's important that requests for each IP operate in a queue manner so that > throttling by IP is effective across all of the processors. And the > throttling needs to be based on the time each request is processed, not > when > the requests are received. > > At the root of all paths to a solution coming to mind is the following: all > requests for one IP address must be queued in a manner that keeps them > separate from all requests for other IP addresses so that intentionally > throttling one IP does not unintentionally throttle any others. > > Ideas are coming to mind, but nothing solid. I'll let you know if anything > comes to mind. Please let me know if I misunderstood your use-case. > > > > -- > View this message in context: > http://activemq.2283324.n4.nabble.com/Marking-a-destination-read-only-tp4691065p4691188.html > Sent from the ActiveMQ - User mailing list archive at Nabble.com. > -- Founder/CEO Spinn3r.com Location: *San Francisco, CA* blog: http://burtonator.wordpress.com … or check out my Google+ profile <https://plus.google.com/102718274791889610666/posts> <http://spinn3r.com>
