Several Clusters on Same Machines -- Efficiently?

We want to run several clusters on the same machines. We are not
          doing this to gain efficiency, since that will obviously suffer. But
          we would like this to be as efficient as possible.
          Briefly, this is the rationale for this setup. Each development group
          would have its own "application cluster" to run servlets, JSPs and
          EJBs. Queries to a legacy entitlements database through a stateless
          session EJB called EntitleBean determine if the user login (basic HTTP
          authentication) is enabled for the requested functionality. By
          running the EntitleBean in a separate "entitlements cluster," we can
          limit the number of connection pools to the entitlements database.
          Suppose our bank of WebLogic servers has M machines, and N application
          clusters run on those machines. Running the EntitleBean in each
          application cluster will require M*N connection pools. A separate
          entitlements cluster will require only M connection pools.
          Question: Can we get something similar to "collocation" optimization?
          I realize that we can't get the equivalence of a method call, and that
          we'll require serialization. Can we avoid a network hit? We would
          like to use a "shortest path" algorithm instead of round-robin, so
          that replica-aware remote stubs would attempt to keep a method call on
          the same machine, although in a different cluster.
          Thanks for any advise,
          David Crane
          Thomson Financial Municipals Group

Hello,
On Windows, Reports Will use the Printer Driver to format the output.
Do you have a Default printer defined ?
Is it a printer similar to the one defined on your "old PC" ?
Regards

Similar Messages

Two clusters on same machine using well know addresses - is it possible?

Hi,
I try to setup two clusters (each with two nodes) on the same machine and want to use WKA for both of them. For the first one I do not have problems and it starts normaly.
When I try to start a node from the second cluster I get this message :
2011-07-07 12:00:06.984/10.710 Oracle Coherence EE 3.5.2/463p5 <D5> (thread=Cluster, member=n/a): Service Cluster joined the cluster with senior servi
ce member n/a
2011-07-07 12:00:14.899/18.625 Oracle Coherence EE 3.5.2/463p5 <Error> (thread=Cluster, member=n/a): This member could not join the cluster because of
a configuration mismatch between this member and the configuration being used by the rest of the cluster. This member specified a cluster name of "Ac
meCluster2" which did not match the name of the running cluster. This indicates that there are multiple clusters on this network attempting to use ove
rlapping network configurations. Rejected by Member(Id=1, Timestamp=2011-07-07 10:08:34.426, Address=<IPaddress>:8088, MachineId=3377, Location=site:c
orproot.net,machine:<MachineName>,process:564, Role=TibcoCepContainerBEMain).
2011-07-07 12:00:14.899/18.625 Oracle Coherence EE 3.5.2/463p5 <D5> (thread=Cluster, member=n/a): Service Cluster left the cluster
2011-07-07 12:00:14.899/18.625 Oracle Coherence EE 3.5.2/463p5 <Error> (thread=main, member=n/a): Error while starting cluster: java.lang.RuntimeExcep
tion: Failed to start Service "Cluster" (ServiceState=SERVICE_STOPPED, STATE_JOINING)
I have configured the "java.property.tangosol.coherence.wka.port" and "java.property.tangosol.coherence.wka1.port" to different values for the second cluster than the values of the first.
Fist of all is such setup possible?
Any idea why I get this error and what else I need to change in the configuration for the second cluster?
greetings Georgi

Hi,
Now it works for me. I think I was setting the WKA on too many places as the documention of the application I have is a little bit confusing.
First I have posted the WKA with java.property as it was specified in the documentation of my application to set them in the script that starts it. I have removed them from the script and used only the xml file to configure coherence.
I am not using the override file but rather change the WKA configuration in the jar file itself. For the real enviornment and not my laptop I will use it.
Also there were two coherence.jar files in my class path and the first one was common for the both clusters and the second one was different for each cluster. In this second one I was changeing tangosol-coherence.xml the logs were showing that the configuration loaded from and also in the classpath this second was before the other one . Maybe I was wrong so I have removed the first coherence.jar that was common.
Maybe the combination of those two things were the problem but now it works. Thanks for your comments.
greetings Georgi

Multiple Instances on the same machine: Several IP Addresses

          Hi all, I want to set up several instances of WL on the
          same server. These instances will be clustered. I am
          using Apache for the WebServer. Here the configuration
          I used: I first change the weblogic.conf file in Apache
          to set up the bridge.I restart apache using apachectl.
          Then, on the WebLogic site, I created a new ipadress
          using ifconfig hme0:1 ip adress ... I created a file
          hostname.hme0:1 in /etc I added this Ip address in /etc/hosts.
          I added this Ip address in the mycluster DNS. I wanted
          to start both WL manually using startWebLogic.sh file
          to see the log file. So, I first assign hostname to WL1.
          Then start WL. I opened an other terminal. Assign the
          hostname to WL2. Then start using startWebLogic.sh. The
          first instance started correctly.When the second started,
          there was an error: when listening to the port 7001,
          this port was already used. That was the same with the
          port 7002.
          Then , my question is How is it possible to configure
          several instances on the same machine using several ip
          addresses and one port. Thank you for your help. Erwan


          Please try -Dweblogic.system.bindAddr=<a.b.c.d> in your startWebLogic.sh.
          Brian
          "Merg" <[email protected]> wrote:
          >> Then , my question is How is it possible to configure
          >> several instances on the same machine using several
          >ip
          >> addresses and one port. Thank you for your help. Eourwan
          >
          >On Win2K/WinNT you can assign several IP adresses to you
          >NIC. (Advanced
          >TCP/IP options.) Each server instance has to run on it's
          >own port.
          >
          >Regards,
          >Merg
          >
          >

Log4j in multiple clusters in the same machine

Hi,
i use log4j and i'm having some problems with my log files. I have two clusters in the same machine. How can i have the logs independent of each other. Now i have one cluster jamming the log of the other cluster. I want this doesn't happen, i want they log separately. How can i do it?
Many thanks in advance,
Nuno

nfalmeida wrote:
Hi,
i use log4j and i'm having some problems with my log files. I have two clusters in the same machine. How can i have the logs independent of each other. Now i have one cluster jamming the log of the other cluster. I want this doesn't happen, i want they log separately. How can i do it?
I hope you have configuration files that configure log4j? The applications/clusters shouldn't use the same configuration.

Running several instances of BPEL PM on the same machine ?

Hi,
What is the best setup to run several conccurent instances of BPEL PM on the same machine (Middle Tier install) ? Of course I would like to minimize the number of components I would need to install.
- Do I need to install several Oracle AS instances (1 Oracle AS = 1 Oracle Home for each BPEL PM instance) ?
- Do I need several Oracle databases for teh dehydration ? (apparently not : http://clemensblog.blogspot.com/2006/02/using-oracle-xe-for-bpel-dehydration.html)
Thanks

Hi unfortunately it is NOT supported to run more then one container with BPEL / and the isntaller will refuse to install inta a midtier that already has BPELPM installed ..
w/o is more j2ee webcache installs (they are basically just the container, em and webcache) and connect them to the DB (different schemas if needed, for failover against the same)
hth clemens

WLS: Several ManagedServer without clustering and same activities

Hi all,
I have a request for special solution for several ManagedServer without clustering and same activities (e.g. SOA or OSB).
In classic case, we have for example 1 AdminServer + 1 ManagedServer (SOA, or OSB). If we want to increase number of SOA-Managedserver, then we add other managed server in WLS-cluster.
In this case, we want to have several "Independent" ManagedServer with different port number and without cluaster but in same domain.
In addition, because we don't want to impact any project in one managed server to other managed server, we need to define different URL:ports for administration: For example: OSB has following URL:
http://localhost:7001/sbconsole
Is it possible to define other ports for other OSB-managed server? e.g. http://localhost:7003/sbconsole, http://localhost:7004/sbconsole
I would like to know, if any experience is available regarding this matter.
Thanks inadvance,
Moh

Hi Moh
1. There are few things that should be clear about the Weblogic Domain and Weblogic Cluster Architecture.
2. Any Weblogic Domain whether its Clustered or Non Clustered will have ONLY ONE Single AdminServer. You CANNOT have more than one AdminServer. Now for Cluster, you can have "n" number of Managed Servers and "n" number of Clusters. Each Cluster is just like a logical grouping of some managed servers. BUT main thing is, if you want to have High Availability architecture to support load balancing, fail over etc, you should and must have a Cluster with managed servers. And expose this Clustered Weblogic Servers into some sort of load balancer like Apace WebServer. Then expose single Apache Web Server URL (host, port + app url) to outside world. General practice is for HA architecture, you do not expose individual managed servers.
3. So coming to your requirement, you can have "n" number of managed servers. And each managed server can deploy their own specific application. And these managed servers naturally will have different port numbers, if they are all on the same physical box. If they are on different boxes, there is no clash of port numbers. Then in front end apache web server, check for each in coming request url and according direct that to appropriate backend weblogic server. This is good but you will NOT get any HA features going in this way. Instead of exposing individual managed servers, I would recommend just create a cluster for each one of them. And expose cluster. For now each cluster can have only 1 manged severs. Later on you can add more also. This gives you HA features. Something like this for example:
Domain = Single Domain
AdminServer = 1 AdminServer only. Not possible more then one.
ManagedServers = 4 (MS1, MS2, MS3, MS4)
Cluster1 = MS1, MS2 -> Deploy App1
Cluster2 = MS3 -> Deploy App2
Cluster3 = MS4 -> Deploy App3
Have a SINGLE Apache WebServer or any other Hard ware load balancer (Apache is free and widely used). Expose this host, port to outside world or actual users (outside or inside). In Apache httpd.conf file, configure like 3 settings, to check for incoming urls and redirect to back end cluster 1 or 2 or 3. Refer docs for full details.
Requests coming to Apache WebServer can use incoming url pattern and pass the request to appropriate backend cluster
http://apachehost:apachport/myApp1 -> Redirect to Cluster1
http://apachehost:apachport/myApp2 -> Redirect to Cluster2
http://apachehost:apachport/myApp3 -> Redirect to Cluster3
Now this gives you full flexibility to add or remove more servers from any of above Clusters. Even if you have single managed server, try to put that into cluster and expose the cluster.
4. Now comes the Adminstration applications like console, sbconsole, em (Enterprise Manager for SOA), cs (Content Server for ECM) etc. All these are deployed only to AdminServer and it makes sense also. Say in above architecture only AdminServer can have visibility to all the managed servers and clusters. Each managed server may not know about the others. Anyhow, the point is, for any Domain you will have only 1 AdminServer and most of the adminstration applications are deployed only to AdminServer. This means literally you will have just one url to control all the applications across all the clusters like http://adminhost:adminport/sbconsole. I am not familiar with sbconole, but for em I am sure its only targetted to admin server. Now, if you manually target these apps to cluster, not sure if this works also. Mostly it may not.
Also, admin urls are not exposed to any users. Only Adminstrators can access them. You can refine some rules like have very few admins with full privileges. Some can do only deployments, some can do only monitoring (read only access) etc.
Thanks
Ravi Jegga

Runing BO 6.5 and Xi 3.0 on the same machine

I have installed 6.5 and Xi 3.0 on the same workstation. Both versions run fine when I run them one at a time (open one, work on it, then exit before starting the other).
However, when I run 6.5 and Xi 3.0 simultaneously, then the next time I try to run either of them, I receive a message asking me to install again.
Has anyone run into this?
Michael O

Hello Constantino,
Is there no way this will work? As I have several customers thinking about or planning an upgrade/migration to XI 3.0, running legacy and new software side-by-side on the same machine is a very important issue, especially as migrations are usually an incremental process, requiring both versions to co-exist for some time.
I have never had any problems running 5.x / 6.x and XI r2 on the same machine, even simultaneous, that's why I'm so surprised at the sudden change.
Any input would be welcome! Thank you!
Kind regards,
Kristof Speeckaert

Tomcat : Hlow to run more than 1 instance of tomcat server on same machine?

Hi,
How to run more than 1 instance of tomcat server on the same machine.
I use tomcat 5.5. ,updated port number for second instance and tried to run,it takes the same old port number .No batch file available with this version to update the home directory to second instance to run the second server..
How to work on clustering and load balancing on tomcat server?
Any inputs?
Thanks in advance..
Ani

Anitha123 wrote:
We have only 1 dedicated server maintained by third party for hosting..
If are only getting your site hosted, then, normally, the host would handle clustering (to your sepcifications, of course, but it is normally the hosts responsibility).
If i need to implement clustering and load balancing for tomcat server then do we need to go for 2 different servers for production environment?
Yes, you would need two, or more, servers.
Also please guide me on how to implement clustering and load balancing for tomcat server by running 2 instances on 2 different machine? I'm a newbie to it..
It should have no (or at least minimal) impact on the code (as long as you are using beans, sessions, and the other components properly). It is handled by the web container (by the tomcat in this case). The tomcat documentation should show how to set it up (if it supports it all, I don't know, we don't have Tomcat at work, and at home I use only a single).

Multiple instances of OC4J on the same machine

Is there a way to cofigure OC4J to have a few processes on the same machine, all listening to the same ports?
My purpose is to develope and deploy a few applications on the same machine, and each application should has its own process, so if onw fails or restarts, it will not effect the other applications and the other programmers.
I don't want to make each application to be bounded to differnt port.
is it possible ?

Two instances of oc4j cannot listen to the same port...you will
get an error.
To do this, you will need to cluster oc4j. This is fairly easy,
but you app needs to be "clusterable". All classes need to be
serializable which are used by your servlets. The
<distributable /> tag needs to be in your web.xml file. No
stateful session beans can be used. (sorry, but oc4j doens't yet
support clustering with sfsb's).
The second step is that your default-web-site.xml or *-web-
site.xml file should be set up for clustering. Do this by
creating a cluster-island attribute in your web-site tag. Also
change the port from 8000 to something like 8080 or 8090 or
something else. It should look like this:
<web-site host="192.168.1.101" port="8080" display-name="some-
name" cluster-island="1">
<default-web-app application="someapp" name="someappname" />
<frontend host="192.168.1.101" port="8000" />
<access-log path="../log/some-web-access.log" />
</web-site>
Your server.xml and application.xml will be the same as when you
don't cluster.
Make sure you have 3 different copies of your j2ee/home
directories...of course with different names....like home1 and
cluster.
Edit the rmi.xml and jms.xml files to change the port number
that the home1 oc4j will listen to. The jms port is usually 9127
by default, so use 9128. The rmi port is 23791, so change it to
21792. This will keep the rmi and jms servers from conflicting.
Make sure that the home1 config/default-web-site.xml uses a
different port than 8080, say 8090, but the cluster-island name
is still 1.
Open terminal windows for each instance of oc4j, and start them
in the normal way.... java -jar orion.jar
Open a terminal window for the j2ee/cluster, and issue this
command:
java -jar loadbalancer.jar -host 192.168.1.101 -port 8000
You should see that each of your oc4j instances are recognized.
There...that wasn't so bad, was it?
regards,
the elephantwalker
www.elephantwalker.com

JMS server targeted to several clusters

Hello,
I am facing a problem with my JMS configuration in weblogic (version 10.3.4.0). I have two clusters. Following the [JMS best practice|http://forums.oracle.com/forums/message.jspa?messageID=9455261#9455261] I have created a JMS server in each Managed server within the clusters. I have created only one JMS Module with a subdeployment. The JMS Module is targeted to all managed servers within both clusters, and the subdeployment is targeted to all JMS servers within the clusters.
At the JMS module I have created a connection Factory targeted to the subdeployment, till there everything is ok.
Afterwards I try to create a uniform distributed queue (UDQ) targeted to the subdeployment and WebLogic throws the following error:
A UDD must be targeted to servers within a single cluster or a single stand-alone server, rather than Cluster ClusterA and Cluster ClusterB
The only solution that I found is to create one JMS Module-subdeployment per JMS Server. Is there a way to have just a JMS Module and use it in all the managed server in several clusters? or I have to set up the same parameters (ConnectionFactories + queue +quota) in each JMS Module.
Thank you very much in advance.
Cheers,
Oscar

Hi Oscar,
The correct way to step-up a UDQ is something like below
C-1 => MS-1 + MS-2
JMSServer-1 => MS-1
JMSServer-2 => MS-2
JMSModule-1 => C-1
Sub_UDQ => JMSServer-1 + JMSServer-2
UDQ => Sub_UDQ
CF => C-1
Where:
======
MS-x are the managed servers
C-x are the Cluster
UDQ is the Uniform Distributed Queue
Sub_UDQ is the Sub-Deployment for UDQ
CF is the Connection Factory
*=>* mean Targeted.
Also to get more information on it you can have a look at the below link which has the step by step instructions which would surely help you out
Topic: Steps to Configure Uniform Distributed Queue (UDQ) on Weblogic Server
http://middlewaremagic.com/weblogic/?p=3747
Regards,
Ravish Mody

Several operation in one machine

Dear All,
my clile requirement is they have one machine they produce several things for example there is mixing machine where they mix chocolate, nuts, biscuits, etc many things in one machine with different capacity how can i map this scenario in sap.
kindly help me in this regard i will be very obliged.
thanks in advance
irfan

Dear,
You need to assign the same work center (Machine) in routing of the material chocolate, nuts, biscuits and define there operion time accordingly. So you can carry out the several operation in one machine .
Hope clear to you.
Regards,
R.Brahmankar

Two running instances of Weblogic on the same machine

I need to run at least two instances of Weblogic on the same machine. My Company GE Capital purchases the Clustered version and the BEA Engineer who did the install did not know Unix and could not intall the instances. Could you please give pointers to on how to do this? You can also write me to [email protected]
Thanks,
John G

....plus, you have to consider the license issue.
Vidar wrote:
>
It does not sound to bad, but would it not be hard to cooperate with other
people, since you will often use API's and EJB's developed and deployed by
other team-members?
Does anybody have any documents or ideas on what the best way to cooperate
in a development-team is?
"Michael Girdley" <[email protected]> skrev i melding
news:39e15942$[email protected]..
Yes, that is the best way to do it.
In practice, we have found that each developer having their ownworkstation
with an instance of WLS works best.
Thanks,
Michael
Michael Girdley
BEA Systems Inc
"newbie" <[email protected]> wrote in message
news:[email protected]..
I am also pondering on the same problem right now...
I am trying to get an environment setup so that multiple
developers can have their own weblogic "instances" to play
with - so that each of them can start/stop the server,
deploy EJBS etc without affecting one another in a UNIX
environment. to accomplish this - do I just install the
server multiple times with different port numbers on the
same machine or is there a a better way to do this?
Appreciate any input.
Thanks
"john Gongora" <[email protected]> wrote:
I need to run at least two instances of Weblogic on the same machine.
My
Company GE Capital purchases the Clustered version and the BEA Engineerwho
did the install did not know Unix and could not intall the instances.Could
you please give pointers to on how to do this? You can also write me to
[email protected]
Thanks,
John G

Limit FPS of multiple instances of vmware on the same machine not over IP

Issue: How can I limit the fps of multiple vmachines on the same machine? (not over IP)
I want to run several vmachines on the same physical machine and limit their fps. I'm aware it can be done with PCoIP but this is not over the network.
This is for non commercial use.
Thanks.

          Please try -Dweblogic.system.bindAddr=<a.b.c.d> in your startWebLogic.sh.
          Brian
          "Merg" <[email protected]> wrote:
          >> Then , my question is How is it possible to configure
          >> several instances on the same machine using several
          >ip
          >> addresses and one port. Thank you for your help. Eourwan
          >
          >On Win2K/WinNT you can assign several IP adresses to you
          >NIC. (Advanced
          >TCP/IP options.) Each server instance has to run on it's
          >own port.
          >
          >Regards,
          >Merg
          >
          >

Running Weblogic 8.1 SP3 and IPlanet 6.0 on same machine

I have requirement to run Weblogic Platform SP3 and IPlanet 6.0 servers on same Solaris machine
for a client of ours.
Please let me if we can do this. What kind of issues we may run into.
Thanks,
Sanjay

Hi
Of course, you can install both at the same machine.
Only remember the memory required ... and the real memory available ... this for don't have swap problems.
Remember too if you are planning to create WLS clusters ... BEA recommends 1 cluster per CPU ... and reserve 1 CPU for the iPlanet web Server. Even, you can install all clusters and web server as your machine support ( under load, try to mantain the CPU usage under 70-80%).
Jin

Interprocess Communication on same machine (Help needed)

Hello everyone, I have a question that I thought you guys could help. Actually I have multiple processes running on a same machine and I want to put messages from one process to the queue of other process what is the best way to achieve this functionality in Java?
Currently I'm working on C++(linux). I have a webservice and some TCP listener processes running on the same machine, and there is another process (i call it processing engine) running one the same machine. So everytime I receive the message from webservice or from TCP socket, I write those messages to the Queue of Processing Engine, and after processing that message my processing engine writes the reply to the queue of webservice/TCP (from wherever it received the message).
I want to achive the same functionality in java. My messages are simple literals/string and I'll write all my processes (webservice/tcp/proc. engine etc) in java aswel.
Secondly I want to process the messages according to FIFO.
Your help will be appreciated.
Thank you.
Edited by: Nafri on Dec 3, 2009 11:02 AM

Rahil.Ali wrote:
What we are trying to accomplish is implement a transaction processing system...
There will several processes working independently, but these processes will be sending messages to one another.. for example
- a TCP process will read the message and translate it into our internal message format and send that formatted message to Processing engine..
- TCP process will then wait for other message from the network..
- Processing Engine will do its job and reply back to TCP process...
We cannot incorporate all the business logic in one process .. since its too big...
Why not? The first server should not involve that much processing. So only the second is the problem. And that is one server.
And what is the actual performance metrics for a single server?
For comparisons I usually use max 2000 txns a second. If you have a server that handles 100 txns a second then you need between 25-40 servers depending on the target box load and back up servers. That really isn't a lot of servers if you really are doing 2000 txns a second (and getting a reasonable per txn fee.)
And I speak from some experience since I have been working on txn processing systems for years.
But I can see the allure of thinking that the front end and back end make more sense. One real advantage is adding another front end (different protocol) without touching the current systems.
Communications options that do work that way would normally be
- messages via sockets
- message queues.
Remember JMS is an API not a solution so you for the second above you need to investigate solutions.
I consider message queues more complex than sockets (message queues of course use sockets and provide more functionality so that is probably a reasonable view.) There have however been some impressive processing speeds reported for them. And getting that same throughput using sockets takes some work.
For either solution spend some time building some non-trivial test setups to gauge real speed paying attention to what you expect your real production needs to be. Experience speaking there since what they claim and what one can actually do for a particular situation can vary.
Keep in mind that a two server solution where you anticipate that the second server might be doing more work might suggest that you need to consider a cluster of second servers servicing the front end server. Either solution requires thought about that before implemetation proceeds.

Several Clusters on Same Machines -- Efficiently?

Similar Messages

Maybe you are looking for