Large messages are one way. The broker will only refuse to accept a message if the store is above the high water mark, not if the message will push it above the high water mark, so if it's at 69% and you enqueue a very large message that's 31% of the limit, you'd get to 100% (and the store would then refuse to accept more messages).
Another is a bug in the code that computes the stats, and there have been several of those over the years. Offhand I don't remember which versions they were fixed in, but you could go through the relevant JIRA issues to see if maybe an upgrade would help. Tim On Jan 15, 2018 3:44 PM, "alainkr" <akreienb...@gmail.com> wrote: > Hello, > > Using activemq 5.10.0 we had a production issue where Memory percent used > was 100% > > > I've beeing trying to reproduce but I can't get the memory used % above the > default cursorMemoryHighWaterMark of 70 % > > I've trying "flooding" several queues with huge messages but I can't get > passed 74% memory usage. > I've tried to start several consumer with the default prefetch of 1000 > which > I thought was going to bring back a lot of message back in memory but no : > memory remains at 74% > > So under what circumstance can the Memory Used approach 100% like we had in > production ? > > Thanks for your help > > Alain > > > > > -- > Sent from: http://activemq.2283324.n4.nabble.com/ActiveMQ-User- > f2341805.html >
