> On many of my high traffic queues Delivering size, Durable delivering size, and Durable persistent size appear to be large negative numbers like -188160082, even while the count is zero. Is that something to be concerned about, or something that I can reset to correct again?
Counts like this should never be negative. Whether you should be "concerned" or not depends on if you rely on these counts to monitor your application. Please open a Jira for this issue and provide a way to reproduce the behavior you're seeing, if possible. > In a queue I'm looking at Acknowledge attempts matches the number of Messages added, but Messages acknowledged is a smaller number. The difference appears to be the number of Messages expired. What does that first statistic mean? "Acknowledge Attempts" tracks the number of times an attempt was made to acknowledge a message whether or not that acknowledgement actually succeeded or not. For example if a client started a transaction that involved acknowledging a message but then later rolled back that transaction the "Acknowledge Attempts" would increment but the "Messages Acknowledged" would not. > If I can reproduce some of these scenarios is this a bug to be reported, and is there any consequence beyond the metrics? Yes, this is a bug. It's not known at this point if there are any consequences aside from the metrics themselves. > Finally, there's an operation resetAllMessageCounters() - does that just put all metrics at 0, so any in flight or queued messages end up reporting negative numbers again, or does it somehow trigger fresh calculations for the stats? The resetAllMessageCounters() method applies specifically to message counters [1]. It doesn't apply generally to the metrics tracked by the individual queue & address MBeans. Justin [1] https://activemq.apache.org/components/artemis/documentation/latest/management.html#message-counters On Tue, Mar 22, 2022 at 1:21 PM Stephen Baker < stephen.ba...@rmssoftwareinc.com> wrote: > I have a number of related questions regarding the various message count > attributes available. > > On many of my high traffic queues Delivering size, Durable delivering > size, and Durable persistent size appear to be large negative numbers like > -188160082, even while the count is zero. Is that something to be concerned > about, or something that I can reset to correct again? > > In a queue I'm looking at Acknowledge attempts matches the number of > Messages added, but Messages acknowledged is a smaller number. The > difference appears to be the number of Messages expired. What does that > first statistic mean? > > Some of my queues have a Delivering Count less than 0, for example my DLQ > which I have purged in the past (possibly from the other side of the > mirror) has a Delivering count of -1989. A few other queues with negative > counts are known to have slow consumers, and finally one with a core bridge > has a negative count. If I can reproduce some of these scenarios is this a > bug to be reported, and is there any consequence beyond the metrics? > > Finally, there's an operation resetAllMessageCounters() - does that just > put all metrics at 0, so any in flight or queued messages end up reporting > negative numbers again, or does it somehow trigger fresh calculations for > the stats? > > Stephen E. Baker > > > >
