| time |
nick |
message |
05:33 |
<forrest> |
hi |
05:56 |
<gigerl> |
hi forest |
05:56 |
<gigerl> |
o he quit already |
13:57 |
<chirino_m> |
rajdavies: wow that xindice stuff is kinda crap.. it does not seem to handle concurrency very well.. |
13:58 |
<chirino_m> |
what a letdown. |
13:59 |
<rajdavies> |
chirino_m: I remember looking at it over a year ago - shame it hasn't got any better :( |
13:59 |
<rajdavies> |
chirino_m: there is a b-tree index implementation - in Kaha - but its very slow :( |
14:00 |
<sbordet> |
hi ! I've been asked how many connections can a broker sustain, say topic consumers; you guys have any rough figure, by chance ? |
14:01 |
<rajdavies> |
sbordet: depends on the platform - but 1000+ |
14:01 |
<sbordet> |
rajdavies: that's enough :) |
14:02 |
<sbordet> |
thanks ! |
14:02 |
<rajdavies> |
np! |
14:08 |
<st3fan> |
i'm still seeing negative QueueSize values in 5.0 with JMX :-( |
14:09 |
<st3fan> |
even if I call the ResetStatistics operation |
14:18 |
<chirino_m> |
rajdavies: you got any benchmark tests setup to compare your btree to the hash? |
14:21 |
<rajdavies> |
just run the test cases |
14:21 |
<rajdavies> |
for hash indexes |
14:23 |
<rajdavies> |
chirino_m: also I don't think the b-tree implementation works :( |
14:23 |
<chirino_m> |
so nothing to compare them? |
14:23 |
<chirino_m> |
ah too bad. |
14:24 |
<chirino_m> |
one day we should make kaha it's own module.. |
14:25 |
<rajdavies> |
chirino_m: agreed! |
14:26 |
<rajdavies> |
chirino_m: though we'd need to make it recoverable |
14:27 |
<chirino_m> |
aye.. |
14:28 |
<chirino_m> |
Just wondering if it was decoupled and we advertised it as a bdb killer if folks would come and help fix it.. |
14:28 |
<jstrachan> |
agreed |
14:29 |
<jstrachan> |
wonder if hadoop has any good btrees? they must use some somewhere? |
14:32 |
<chirino_m> |
jstrachan: peeking right now.. |
14:49 |
<sbordet> |
rajdavies: while browsing amq code, I found this (BrokerService#711): |
14:49 |
<sbordet> |
systemUsage.getTempUsage().setLimit(1024 * 1024 * 1024 * 100); // 10 Gb |
14:50 |
<sbordet> |
however, 1024 * 1024 * 1024 * 100 overflows int, so actually this seems equivalent to setLimit(0) |
14:51 |
<sbordet> |
should I raise a jira ? |
14:59 |
<rajdavies> |
sbordet: crikey! - yes plese! |
15:00 |
<sbordet> |
it's enough to add * 100L (note the L) to the multiplication :) |
15:01 |
<sbordet> |
well replace that is :) |
15:01 |
<sbordet> |
raising jira now... |
15:02 |
<rajdavies> |
thx! |
15:02 |
<sbordet> |
np |
15:02 |
<sbordet> |
rajdavies: one q: storeUsage setting is per broker, right ? |
15:02 |
<rajdavies> |
yep - correct |
15:02 |
<sbordet> |
thx ! |
15:09 |
<sbordet> |
rajdavies: https://issues.apache.org/activemq/browse/AMQ-1858 |
15:10 |
<rajdavies> |
thx! |
15:21 |
<st3fan> |
hmm i'm doing 200-300 messages per second |
15:22 |
<st3fan> |
i have a whole bunch of interesting exceptions in my activemq.log for 5.1.0 btw |
15:22 |
<st3fan> |
NPEs, but also things like PropertyNotFoundException and 'Failed to build body from bytes' .. should I report things like that? |
15:29 |
<sbordet> |
I am sending a JMS message from Ruby (using stomp) to a topic |
15:30 |
<sbordet> |
I receive the message in a Java application, but the said message is always a BytesMessage |
15:31 |
<sbordet> |
is there a way to have the java consumer to receive a TextMessage ? |
15:37 |
<rajdavies> |
yep - please send in a jira |
