Gary,
It does appear that it is just the stats that are wrong in this case. The
messages
are getting delivered as expected (as long as there is an application
consumer on
that broker).
A couple of other questions:
• Since I only seem to see this issue when there are 2 or more brokers, is
there
ok, I guess this is going to be very hard to track down without a
reproducible test case. Can you confirm that it is just the stats that
are wrong in this case, and that all expected messages are getting
delivered as expected.
On 9 September 2010 14:28, Colleen Velo wrote:
> Gary,
>
> That doesn'
Gary,
That doesn't seem to be the case here - namely, the JConsole queue count is
accurately
reflecting what the ActiveMQ web console is showing. I have rechecked our
application
code as well as the ActiveMQ server settings and in both cases we are using
the default
*dispatchAsync* value of false
Is it by any chance the inflight count in jconsole that is negative?
There is a good explanation for that,
http://blog.garytully.com/2010/01/activemq-prefetch-and-asyncdispatch.html
On 8 September 2010 21:42, Colleen Velo wrote:
> Gary,
> * *
> *Sorry - last message was incomplete - technical dif
Gary,
* *
*Sorry - last message was incomplete - technical difficulties*
Haven't had any luck reproducing this issue with a JMeter test, however, I
can reproduce
it on a fairly regular basis using our in-house application script, which
puts tons of application
messages on the queues.
Also, I don'
Gary,
Haven't had any luck reproducing this issue with a JMeter test, however, I
can reproduce
it on a fairly regular basis using our in-house application script, which
puts
Also, I don't seem to see this issue when there is only a single broker
running, only when
there is two or more running in
Gary,
Forgot to add that I do see the same thing in JConsole, that is the negative
counts.
On Fri, Sep 3, 2010 at 1:37 PM, Colleen Velo wrote:
> Gary,
>
> Thanks for the clarification - still working on a JUnit test.
> Currently right now the steps that I use to reproduce it is
> to have the 3
Gary,
Thanks for the clarification - still working on a JUnit test.
Currently right now the steps that I use to reproduce it is
to have the 3 producers each send *60,000* messages each to
the ActiveMQ cluster at the same time - more often than
not, when the queue is trending down towards 0 on the
Some series of steps that can reproduce or in the ideal case a junit
test we can drop into the source.
On Wednesday, September 1, 2010, Colleen Velo wrote:
> Gary
> re: JConsole - don't know yet. I will try to hook JConsole up and look
> through that as well -
> currently using the bundled web ad
Gary
re: JConsole - don't know yet. I will try to hook JConsole up and look
through that as well -
currently using the bundled web admin console.
As far as test case, I don't have one - just the standard message traffic
from
our producers to our consumers (slightly less than the traffic that is
ru
does jconsole show the same negative number? Do you have a test case?
On 1 September 2010 13:52, Colleen Velo wrote:
> After only running my 3 ActiveMQ brokers for short time, I am seeing
> negative numbers in
> the "Number of Pending Messages" column. I started seeing this at 5.3.2
> (after upg
After only running my 3 ActiveMQ brokers for short time, I am seeing
negative numbers in
the "Number of Pending Messages" column. I started seeing this at 5.3.2
(after upgrading
from 5.2.0) and opened and issue on it:
• https://issues.apache.org/activemq/browse/AMQ-2887
It was recomme
12 matches
Mail list logo
