to ?
Session (Tomcat, Jetty) ?
EJBs ?
I think replis async by default.
Joao Clemente wrote:
Hi.
(This message is in particular for Bela, but others feedback is welcome)
I'm running some tests and even with a very low request rate (5 req /
minute
= 1 request / 12 sec) I'm seeing my respose
it's
because of a overhead of Jgroups ack'ing the replication before letting
the master node reply to the user.
(question for bela) Can it be?
If so, can I put jgroups in async mode, either delaying the replication
message or not waiting for the ack from replicas?
Thank you
Joao Clemente
haven't timed out? I think the default HTTP session timeout is
15 minutes.
Joao Clemente wrote:
Ok, trying to figure out what is going wrong with my tests, I decided to
do
the following:
I've runned my 100 user test 10x. Test consists on about 15-20 user
actions (http requests
Bela, if you check my thread Cluster replica absurd memory usage I figured
out I am doing a 1.6 requests per second to the cluster. If every request
issued 10-20 interactions between my application components, it yould be
16-32 cluster msgs/sec, way far from being too much I guess...
Thank you but
are created
when using clustering compared to the non-clustered case. Furthermore, it is
up to the JVM GC to decide when to reclaim that memory. Maybe the memory is
used but could be reclaimed.
Joao Clemente wrote:
[snip]
Now, who can explain:
1 - why the 1rst machine memory usage increase? maybe
Bela, could you point me to some documentation or provide a little better
explanation of what to do?
I'm a bit lost here... : I downloaded javagroups 2.1.1, found a fc.xml file
inside (btw, there's a flow-control.xml file aswell, that seems to be a copy
of fc.xml), checked what was inside my
memory usage getting larger and larger each run of
the test? It seems to grow about 60 Mb per run...
--
Joao Clemente - jpcl at rnl ist utl pt
---
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
, btw, I'm using jboss 3.0.8 with j2sdk 1.4.1_01 on windows 2000.
- Original Message -
From: Joao Clemente [EMAIL PROTECTED]
To: [EMAIL PROTECTED]
Sent: Wednesday, September 17, 2003 4:33 AM
Subject: [JBoss-user] Cluster replica absurd memory usage ?!?
Repeated the test directing requests
to
maximum usage, and after hours of not loading the server the memory usage
does not decrease?
- Could I set more aggressive memory settings to java that could free more
memory, making my app work without the out of memory?
--
Joao Clemente - jpcl at rnl ist utl pt
.
Joao Clemente wrote:
Hi! Load testing my application ( with Jboss 3.0.8 ) I had these strange
results:
If using 1 jboss node, with cluster settings active but without another
machine, I runned a test scenario that consisted of 10, 100 and 200 users
navigating through my site.
I had
provide a servername to do jndi queries,
we'll not have HA or LB in that client interactions with the cluster... or
can we?
Thank you
--
Joao Clemente - jpcl at rnl ist utl pt
---
This sf.net email is sponsored by:ThinkGeek
Welcome to geek
Sorry for the last message.Forget it.
I had just read Build optimized Jboss version message and I tough you're
answering to it. I only notisted the mistake after double-checking the
subjects.
Sorry.
---
This sf.net email is sponsored
- Original Message -
From: Sheldon Hearn [EMAIL PROTECTED]
To: [EMAIL PROTECTED]
On (2003/09/01 09:57), Neal Sanche wrote:
Another option would be to investigate Pre-Compiling your JSPs so that
the compilation time won'[t affect the first user on the system.
I've done this with
If the Session bean is in the same JVM as your MDB then it will not be
load balanced. THe logic behind this is that local calls will always
be faster.
Sounds like that is the solution for my problem. But what can i do when
i wish to not deploy my application in one node that is in a
() and concurrent calls on stateful
beans
Probably, but the deadlock thread dump is needed to be sure.
--
Scott Stark
Chief Technology Officer
JBoss Group, LLC
Joao Clemente wrote:
I'm trying to use the following pattern to recover my ejb's
cant you just use ./back_office ? Why not? I particulary am fond of using
the following setup:
./deploy,./myDeploy,../../../myDeploy
This gives me a folder at the same level as deploy that I can use to pu my
own stuff, and a foder at the same level as jboss-xxx to put apps when I'm
testing with
... But this is another
story...
Hope maybe our bad (old) design will help someone to avoid this mistakes :-)
Joao Clemente
- Original Message -
From: Joao Clemente [EMAIL PROTECTED]
Sent: Monday, July 28, 2003 4:36 PM
We're changing our app so that we can recover the transient parts of our
to obtain an
exception (concurrent calls on stateful beans).
Using this pattern with methods other than ejbActivate we get that
exception, so shouldn't it occur here aswell?
ejb1 {
ejbActivate() {
ejb2.someMethod()
}
}
ejb2 {
ejbActivate() {
ejb1.someMethod()
}
}
--
Joao Clemente
a call to EJB1.businessMethod1(), what
is the pattern that allows to recover value2 ?
--
Joao Clemente - jpcl at rnl ist utl pt
---
This SF.Net email sponsored by: Free pre-built ASP.NET sites including
Data Reports, E-commerce, Portals
Ok. I've read the spec and I see no reason why interlaced business-method
calls triggered by ejbActivate()'s should not throw the referred exception.
Using jboss 3.0.8 it locks silently.
--
Joao Clemente - jpcl at rnl ist utl pt
for ejb recovery wanted
Hello Joao,
Are you using demarcation and saving the transaction state with separate
containers?
---Original Message---
From: Joao Clemente [EMAIL PROTECTED]
Sent: 08/01/03 01:11 PM
To: [EMAIL PROTECTED]
Subject: [JBoss-user] Pattern advice for ejb recovery
We're changing our app so that we can recover the transient parts of our
objects so that we can eventually support failover.
One of our recovery methods fails with
Application error: no concurrent calls on stateful beans
and we're being unable to trace what is responsible for this behaviour. We
I'll get you one, but I'm not sure what you want:
- The exception stacktrace?
- The server stack trace (dump) at the time of the exception? This last one
I don't know how to do programatically...
- Original Message -
From: Sacha Labourey [EMAIL PROTECTED]
Sent: Monday, July 28, 2003
I've been using the binaries in a university project wich involves
clustering.
However, I've found and reported bugs/problems that were fixed on cvs. Until
the next release was provided, I had to resort to building from CVS to
benefit from those fixes.
If you don't need something from cvs, I see
Have you seen any problem on the way I'm using replication?
I am suposed to keep the ejb references on the servlet httpSession , right?
- Original Message -
From: Joao Clemente [EMAIL PROTECTED]
To: [EMAIL PROTECTED]
Sent: Wednesday, July 16, 2003 4:22 PM
Subject: Re: [JBoss-user
-
From: Joao Clemente [EMAIL PROTECTED]
To: [EMAIL PROTECTED]
Sent: Tuesday, July 15, 2003 12:41 PM
- Original Message -
From: Scott M Stark [EMAIL PROTECTED]
Subject: Re: [JBoss-user] NoSuchObjectException: Could not activate;
failed
to recover session
Show the full stacktrace
- Original Message -
From: Scott M Stark [EMAIL PROTECTED]
Sent: Wednesday, July 16, 2003 3:11 PM
This shows that the call is not going through the HA aware proxy. It is
going
through the local invoker and bypassing the HA aware proxy because of the
collocation of the caller and
- Original Message -
From: Scott M Stark [EMAIL PROTECTED]
Subject: Re: [JBoss-user] NoSuchObjectException: Could not activate; failed
to recover session
Show the full stacktrace of this error to demonstrate that the request is
in
fact going through an HA capable proxy.
Well, it if
directly to JB1)
But I'm not sure if this makes a diference or not...
- Original Message -
From: Joao Clemente [EMAIL PROTECTED]
To: [EMAIL PROTECTED]
Sent: Sunday, July 13, 2003 8:21 PM
Subject: [JBoss-user] [Cluster test app] fails to recover ejb session
Well, I can't explain why, but I've setup my cluster with the a loadbalancer
in front, instead of interacting directly with one of the jboss servers, and
this way I don't have an exception...
If someone has followed the problem I described here, I have an hipoteses
for what was happening, but I
I'm using JBoss 3.0.8 and testing clustering failover, in particular the
recovery of a ejb session.
I would like to ask if someone as executed such test and has suceeded with
it.
(Strangely I was almost sure I had good results in this test some months
ago, with a previous JBoss version)
I am
The new problem I'm having:
Trying to figure out the maximum number of concurrent connections that my
server deals with, I'm being able to open more than 20 connections at a
time, and less than 30 connections... The strange thing is that this value
only applies to the new connections I'm trying to
]
Sent: Sunday, June 08, 2003 5:22 AM
Subject: RE: [JBoss-user] http session clustering
You're problem is still not resolved Joao?!?
-Original Message-
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of
Joao Clemente
Sent: vendredi, 6. juin 2003 14:48
To: [EMAIL
On Wed, Mar 26, 2003 at 07:45:55PM -0500, Dan K. wrote:
I believe there is a bug with getting the session in JBoss 3.0.6 using the
all configuration with distributable/ in web.xml. The behavior is
that every request to my test servlet gets a new session id (new session
is created).
If I
- On Fri, 09 Aug 2002 14:05:34 -0700
Greg Turner [EMAIL PROTECTED] wrote:
I am looking for a free software solution to load balance http requests
to multiple instances of JBoss/Jetty in a cluster inorder to test
ClusteredHttpSessions. Any ideas? What does everyone
Hi.
As I was looking for a good load balancing solution for JBoss, it just came to my mind
this question:
If httpSession is replicated, all machines have all session instances in memory... So,
there is no need to use a sticky-session load balancing algorithm, right?
Furthermore, as we do not
You need to copy the jbossha-httpsession.sar into the deploy directory. This
file can be found, in one of the docs subfolders.
Sorry but I looked and looked and there is no such file.
Maybe it's only on 3.0.1 RC ? Or maybe in the JbossTomcat version?
I am using 3.0.0 (with Jetty)
--
Put
Did you download just the binaries? If so, go to this URL and download the
source. Then go to build directory and run the build script. Then go to
cluster/output/lib where you will find the sar file.
Yes, I downloaded the binaries... Ok, I'll do that
38 matches
Mail list logo
