<< 16-May-2008 : hausbot on #activemq at codehaus [download] [back] >>
 
 
time nick message

15:13

<[HypE]>

hi all

15:16

<[HypE]>

anyone alive here? :-)

15:18

<chirino_m>

chirino_m check his heart...

15:18

<chirino_m>

yeah Think so.

15:19

<[HypE]>

i'm playing around with 5.1.0 and msgs seem to be delivered more reliably with a prefetch size of 0, i only have one consumer.. any hints on things to checkout?

15:20

<[HypE]>

ttyl of 0 on the msgs too

15:22

<[HypE]>

prefetch size of 1 rather, 0 would not be good

15:23

<chirino_m>

[HypE]: why is that?

15:23

<chirino_m>

what happens when prefetch is > 0

15:23

<[HypE]>

just assuming 0 would not be good, i have it set to 1

15:23

<chirino_m>

well 0 is supported.

15:24

<chirino_m>

and disable prefetching

15:24

<[HypE]>

some messages are just not being picked up by the client

15:24

<chirino_m>

so when you call consumer.receive() then the client does a RPC to the broker to pull a message

15:24

<chirino_m>

much slower than prefetching

15:24

<[HypE]>

k, I'll try with 0 too then.. is there anything special i should be doing on the client side to handle prefetching reliably? i wouldn't think so

15:24

<chirino_m>

that's why most folks don't use it.

15:24

<[HypE]>

yea, gotcha

15:24

<chirino_m>

I don't understand my prefetching is not reliable in your case.

15:25

<chirino_m>

are you using a java client?

15:25

<[HypE]>

i don't either..i don't really know if it's prefetch size per se..but it does seem more reliable with a prefetch of 1

15:25

<[HypE]>

yea, tried with and without MessageListener

15:25

<chirino_m>

so java client right?

15:25

<[HypE]>

yea..java client

15:26

<chirino_m>

can you make a junit test case showing the problem?

15:26

<[HypE]>

yea, i got one, would just need to remove some junk from it

15:27

<chirino_m>

sweet.

15:27

<chirino_m>

if your allowed attach it as a contribution to the ASF in the activemq jira.. that we we can add to the test suite.

15:27

<chirino_m>

if you can't then just email it to me and I'll run it.

15:28

<chirino_m>

jstrachan: back home?

15:28

<[HypE]>

yea, i'd like for someone to review first maybe, maybe i'm just doing something wrong..but i've given it a fair effort

15:28

<jstrachan>

chirino_m: just arrived yeah

15:29

<chirino_m>

[HypE]: yep.

15:40

<chirino_m>

jstrachan: what do you think of http://www.nabble.com/.../Asynchronous-acknowledgements---are-they-correct--td17277656s2354.html#a17277656 ??

15:42

<chirino_m>

jstrachan: are you the owner for the nabble stuff for amq?

15:42

<chirino_m>

wondering if we should not use the embedding option nabble has.

15:52

<jstrachan>

you mean read that thread or look at the layout?

15:53

<chirino_m>

well both

15:53

<chirino_m>

:)

15:55

<chirino_m>

technically I think he might be right.

16:00

<Drizzt321>

I'm having some trouble with my AQ 4.x instance. I'm connecting with 5.1 libs, but everything up until now has seemed to work fine. I was sending lots of data through it to another application (small, TextMessage's, but many message) over night, but when I checked it this morning things seemed hung. The app was still responsive, but ActiveMQ was not

16:00

<Drizzt321>

I can barely connect via JMX, but I don't get ANY data, not even the VM Summary

16:00

<Drizzt321>

and I am getting java.io.IOException: Wire format negotiation timeout: peer did not send his wire format. on both client and server

16:01

<Drizzt321>

It might be related to the broker's memory usage as per activemq-users@geronimo.apache.org/msg05949.html">http://www.mail-archive.com/activemq-users@geronimo.apache.org/msg05949.html

16:01

<Drizzt321>

but I don't have out of memory exceptions

16:02

<Drizzt321>

It the process is using up >12% of my memory (7857 RAM as reported by free -m), and

16:04

<Drizzt321>

anyone have any thoughts on how to get things moving again?

16:10

<chirino_m>

Drizzt321: not sure I would need get my hands on it.

16:11

<chirino_m>

rebooting the broker is not an option?

16:11

<Drizzt321>

chirino_m: the biggest thing, is that I can't even get anything to load up via JMX

16:12

<Drizzt321>

well, it is, I'd have to restart part of the data load

16:12

<Drizzt321>

which isn't too too far along at least

16:12

<chirino_m>

enable debugging on the broker vm

16:12

<Drizzt321>

but it would be really, really cool if I could just get things to start moving

16:12

<chirino_m>

and attach a debugger if things get really bad.

16:12

<Drizzt321>

how do you do that without going through JMX?

Drone v1.4 © 2002-2005 Uwyn RIFE powered