Like I said in my other message, a lot more information would be
necessary to find out exactly what is causing the problem that you
mention.
If it is related to the database connection pool one instance can vary
a lot from another if they are running different code. You may have
some custom code that does not handle transactions properly. With a
more recent version of OFBiz the connection pool will recover bad
connections pretty well, but the old one did not. What that means is
that when the bad code runs it may put a connection in a bad state and
the connection pool will never recover that connection, and eventually
the connection pool will be full of bad connections, and that will
bring down your server. Sometimes those bad connections, or the code
related to them, are related to bad threads that consume a lot of
processor time.
Still, that is just a guess...
-David
On Jul 27, 2009, at 2:44 PM, S K Pradeep Kumar wrote:
Hi,
But i have compered my application with the other similar
applications
runinng on the server having similar configuration but those apps are
running fine the issue is only with this apps. i.e. www.tribu.co.uk
With regards,
S K Pradeep kumar
On Tue, Jul 28, 2009 at 1:59 AM, David E Jones <[email protected]> wrote:
The release4.0 branch has issues that have never been resolved.
Please
understand that each release branch represents a sub-community of the
overall OFBiz community, or in other words there are more people
involved
with OFBiz in general than there are with a particular release
branch. A lot
of things, including low-level things, were improved in the two years
between the release4.0 branch and the release09.04 branch.
It's hard to say exactly what issue you are running into that at
this point
I don't think anyone involved with the development and maintenance
of OFBiz
would recommend using the release4.0 branch because of the issues
it has
(including significant connection pool problems with the old Minerva
connection pool) and also because of the significant new features
and better
ways of doing things and more complete functionality in the
release09.04
branch.
-David
On Jul 27, 2009, at 2:10 PM, S K Pradeep Kumar wrote:
Hi,
I have tested with the following instruction given by Henning, and
compered the results with ofbiz release 4.0 and it seems that my
application
is working fine locally,
But my problem persists online and in my online application i am
having.
Application is running on linux i836
With 2gb of ram
And
my java consumes 30 to 40 % of memory normally and mySQL consumes
7.9 % of
memory.
Some times the CPU usage goes upto 150 % and hangs up the server
and java
consumes more than 60% memory.
Can anybody tell me what might be the reason.?
i have compared with the same such kind of apps deployed on vps they
consumes 2 to 4 % of memory for MySQL but this apps consumes 7.9% of
memory.
Might be it is the problem
If yes please, guide so that i can work on this.
Please...
With regards,
S K Pradeep kumar
On Sun, Jul 26, 2009 at 9:15 PM, Henning <[email protected]>
wrote:
Operating a Java Server on a VPS requires some tweaking. First of
all
you should check the current memory consumption in your scenario
using,
for example, jconsole (which is part of the JDK since 1.5). In
order to
connect to a running jconsole instance, you should some JMX system
properties, e.g.
-Dcom.sun.management.config.file=management.properties
and put a management.properties file in your ofbiz folder (next to
startofbiz.sh) that looks like this:
com.sun.management.jmxremote=true
com.sun.management.jmxremote.port=7777
com.sun.management.jmxremote.ssl=false
com.sun.management.jmxremote.authenticate=false
(assuming you are using the SUN JDKs).
Note: this should never be used in production as the JMX access is
completely unsecured this way!
With the config above, you can connect via
jconsole localhost:7777
Now run your load test or simulate your scenario and check the
memory
data in jconsole. Trigger a few GC's.
Check the various memory regions by going to the "Memory" tab or
directly to MBeans --> java.lang --> MemoryPool.
In my experience in particular the CodeCache pool is very much
oversized
and the SUN JVM allocates rather large chunks of memory here.
Which is
normally not a problem, but VPS's tend to have very tight virtual
memory
quotas (which seems really unnecessary).
Anyway, in order to redruce memory consumption, you should limit
heap,
perm space, code cache, using the VM settings
-Xms<heap>M -Xmx<heap>M -XX:ReservedCodeCacheSize=<cc>m
-XX:MaxPermSize=<perm>m
where <heap>,<cc>,<perm> are placeholders for your settings.
While <perm> and <cc> should not change dramatically after your
scenario
has run (so add a little on top of your measurements), the heap
size
will also have
to reflect user session memory (so add some more. No idea how big a
session in ofbiz gets in the end).
Good luck,
Henning
Am Samstag, den 25.07.2009, 14:51 +0530 schrieb S K Pradeep Kumar:
Hi All,
We have deployed the ofbiz with tomcat on VPS server.
It is creating too many dummy sessions and consuming loot of
memory.
Could you please help in this.
What might be the reason?
How can we customize the application to reduce memory consumption?
--
With regards,
S K Pradeep kumar
--
With regards,
S K Pradeep kumar
