Load balancing Tomcat

Hi all. We have a web application that is load balancing across a number of Unix servers, each with its own instance of Tomcat.
However we wish to install more than one version of Tomcat on a box. We have tried it, but the session data is lost as soon as a request hits the other version. So my questions are:
Is it possible to run more than one version of Tomcat on a box, and for the sessions data to be shared?
Can session data be written by Tomcat to a file which can then be shared by other instances?
Thanks in advance for any help or pointers with this.

I got this example working excellently with two Tomcat instances.
I tried to configure it to work with three - do you know if this is possible? Something didn't work quite right, and the tutorial page only mentions two tomcats, never says if it can be adapted for more than that...
Thanks for advice.

Similar Messages

Load Balancing, Tomcat, and SharePoint

I'm a new BusinessObjects customer and am working on getting all of the hardware in place for a new install. The initial plan is to have two BOE servers, and two tomcat virtual servers, with a hardware based load balancer (F5 BIG-IP Switch: Local Traffic Manager 1600 4GB, possibly) in front of the tomcat servers to handle the load balancing.
But, I'm starting to think that it would be a good idea for us to integrate directly in to our SharePoint portal using the SAP SharePoint Integration option.
So, here's my question. If we do that, then our users will get to their BusinessObjects information via SharePoint. So, SharePoint and IIS will be the web server. Will we still have a need for the Tomcat servers? Will SharePoint point to the load balancer, then to Tomcat, and finally to the BOE server, or will it go right to the BOE server, thus negating the need for the tomcat servers altogether, and also even negating the need for the load balancer? If so, can the BOE app servers themselves still be load balanced?
Hopefully this all makes sense - like I said, we're a new customer, so I don't fully understand all of what the servers are doing. I've spent several days searching the forums & the web & reading documentation and haven't come up with an answer yet, so I'm reaching out to you all & hoping somebody can clear it up for me. Thanks!!

Thank you for your input, Denis, this does help explain things. Also, thanks for pointing me to that Windows Patterns document. I hadn't seen that one - since we're starting out on 4.0, I've only been looking at documents under the 4.0 folders...
It sounds like we may want to utilize this same load balancer to balance the SharePoint traffic as well, while we're at it. Right now our SharePoint server is clustered using Microsoft Clustering, but not load balanced in any way. Or maybe we just post a link from within SharePoint to the regular old BI LaunchPad and call it a day!
So, if I understand correctly, if we utilize SharePoint, any traffic/processing that the Tomcat server would have handled would now be handled by the SharePoint server, which could potentially be significant. The actual processing of the reports, though, will still get handled by the BOE Cluster, which takes care of its own load balancing, so we'll be fine. Really the only thing the hardware load balancer does is allow the presentation layer to be load balanced - the layers beyond that get load balanced automatically via CMS. Is that all somewhat valid? In the patterns doc, there's Application Servers separate from the BOE Servers. We were just going to have two BOE servers and two Web servers. Where do the application servers fall in, and are they load balanced using the hardware piece?
If we were just concerned about High Availability, it seems like maybe we could more easily use Windows Clustering on those Tomcat servers, and avoid the hardware component altogether. Do the Web servers really get hammered that hard that we need them load balanced? It seems like the BOE servers are the ones doing the heavy resource intensive tasks and we get them load balanced without the hardware anyway, so maybe load balancing the web servers is overkill..... I'm sure a lot of this is tough to answer, I'm just trying to get a feel for it. I want us to have our best performance & be somewhat future-proofed, but don't want to buy things that I don't need to! We're probably looking at a small, 10-20 concurrent user setup for starters, and out to say 200 concurrent max once it's fully rolled out...
Thanks again!!

Hardware clustering/load balancing/failover with Tomcat

Hello forum!
I recently bought a Cisco 1801, and it sure is capable! Anyhow, I've got a hobby website that is getting a fair bit of traffic - approaching too much for one node to handle and it's time to start thinking about distributing the load.
I'd like to do a little clustering of server nodes running Apache Geronimo, which is J2EE running atop Apache Tomcat. For the sake of keeping things generic, let's just call it Tomcat because it configures the same way.
I do not run Apache HTTP Server as a proxy, I only run Tomcat directly connected to the internet. I do this for performance reasons.
Anyhow, I'm wondering if any of you evil geniuses could suggest a way that I could cluster Tomcat nodes directly using the router to serve as a hardware load balancer and have the whole sticky session thing with failover, etc... All of the documents I find on the subject discuss clustering by way of Apache HTTP with Mod_JK.
I have already asked this question on the hardware side, and got great information about the capable load balancing features my router sports (but limited compared to Cisco CSS products.)
Now I'm wondering if anyone has experience taking an open source application server like Geronimo or Tomcat or JBoss and clustering it using hardware load balancing. What kinds of Tomcat configurations, if any, do I need to add for things like sticky sessions and failover? Or, is all that automatic?
Thanks so much for reading and for any replies. If there is a better forum for my question, please direct me there.
Cheers,
Dave Woldrich
http://CardMeeting.com

This occurs rarely when the Tomcat process is not able to connect to the database. The database connection problem is an internal cause which manifests externally as missing fields in reports.
Workaround: Restart the Apache process and the Tomcat process. From the CLI on your CiscoWorks Server, enter the following commands in the specified sequence:
1. pdterm Apache
2. pdterm Tomcat
3. pdexec Tomcat
4. pdexec Apache

Tomcat load balancing and frames

Hello !
I am having an web applications built on Servlets and JSP, but in my application I am using frames so after I am connecting to the application I am having one JSP wtih 3 frames (JSP).For the web site I am using sessions.My application is working ok on one Tomcat, but when I tried to put the application on server where the configuration is a loadbalancing for 2 Tomcats I am having problems because frames:after I am connecting to the application I am seeing the first frame a thousand of times (like a cycle).I am using Tomcat 4.x
Can anybody help me with some ideeas?
Thank you!

First off, consider switching to Tomcat 5. The JSP performance alone is worth the switch. It is a much more robust servlet engine than Tomcat 4.
Second, your frame problem is likely due to loadbalancing directing each frame to a different server. If you like that approach, examine your method for HTTP session replication. How are sessions being replicated between Tomcat instances? (Or even are they?) If you want true perfect load balancing you'll need good session replication.
If that doesn't work you may consider a loadbalancing scheme involving server affinity, also known as "sticky sessions." That way all requests for a given session will go to the same server. With this scheme your frames will all render properly because they will be served up by the same machine.
For Tomcat clustering and loadbalancing I recommend Apache with the MOD_JK plugin for tomcat clustering, loadbalancing, and failover. MOD_JK supports both round-robin and sticky session loadbalancing.

Tomcat load balancing

I'm going to develop a load balanced system by using tomcat as my servers. i would like to ask is there any way i can integrate my own load balance java program with the tomcat server? i'm not going to use the worker property that is prepared by the tomcat....

I'm not aware what the worker property is, but doing simple load balancing (round-robin, load-based) via the proxy mechanism is quite fair.
It involves having a front-end servlet that accepts all connections, maintains it's own session with the browser that keeps that session bound (mapped) to a server FOO. When the request comes in, just proxy the request over to FOO; if a new request/new session comes in, the next server in line is BAR, establish a new session, map BAR to that session, and reread paragraph ;-)

TOMCAT SERVER CONNECTED TO LOAD BALANCER

Hi ,
scenario:
Tomcat server addedd to the load balancer for performance .
Q&A:
We have added tomcat(new server) to load balancer,now all the services are up and running,but i want to know the validation of how the new server added to the cluster will have the connections via load balancer to this server. is there any simple way where by we can go by looking at some logs, or load, or connection statistics for the Tomcat server whilst users connect.
please gurus can you help,
1.HOW TO VALIDATE THE LB IS WORKING FINE .
2.TOMCAT SERVICES VIA LB (CONNECTION)
3.OTHER VALIDATION TO BE CONSIDERED.
Regards,

SAP Business Objects really cannot verify if your load balancer is working fine, you should take that up with your load balancer manufacturer or IT folks that manage it.
When using a load balancer with business objects deployment on tomcat you should enable session persistance (this is mandatory in XIR2) and may be needed in some circumstances in XI 3.x.
A little trick I've seen customers do is to change an image file in the deployment to have a #1 or 2 or 3 or... depending on the tomcat server, this way they could tell which server the load balancer was redirecting to. To do this right click the image when accessing tomcat (to get it's location in the deployment) and simply edit with MS paint or favorite image editor (back up the original 1st just in case...)
Regards,
Tim

CF 10 Load-Balancing with Remote Instances

I was reading an article on Clustering/LB/HA using CF8, but have not found any updates for CF10.
Using VM VirtualBox to setup a few virtual servers, I am looking to setup a load balancing of ColdFusion 10 on 2 remote instances. The goal would be have ColdFusion Cluster Manager be able to point http request to one of the two servers based on load/availability. Not really having a hardware cluster/failover setup, just managing resources on two CF instances instead of a standalone.
The servers are Windows Server 2008 R2 with IIS7.5 and ColdFusion 10 Enterprise on installed on 3 of these machines. Let's call them CF-LBManager, CF-Web1, and CF-Web 2. In the CF Docs, they show the Cluster Manager adding the local CF instance and "if you want" a remote instance. However, this scenario would require the main instance to be running and not fail for it to direct to the other instance.
I am trying to set this up now with CF-LBManager as just a manager of the requests coming in. In the Enterprise Manager >> Instance Manager, the local instance is shown and I add the two remote instances with the correct Remote Port, JVM Route, etc. I also made sure the <Cluster>...</Cluster> block was added to the two remote instances (CF-Web1 and CF-Web2) \runtime\conf\server.xml file too, Jetty Services also is running. Now under the Enterprise Manager >> Cluster Manager I add the two remote instances to the cluster, not the local instance on CF-LBManager with Multicast Port and Sticky Sessions enabled. On Submit, I get a green message "You must restart all the server instances and any configured webservers for these changes to take effect.". I go ahead and reboot the servers and come back.
I now browse to the ColdFusion page as a test on CF-Web1 and CF-Web2 to make sure CF is running properly, they do. I then browse the IP of the CF-LBManager, however it only returns the local IIS web site and not redirect to one of the two cluster members. I am not seeing any message on the coldfusion-out.log on the remote instances. Am I not setting this up correctly or not enabling the Cluster Manager to take over and pass along the requests to those in the cluster?

Unfortunatley I don't have a lot of experience with CF10 on Windows, but if you are running CF behind IIS I think you will need to update the Tomcat connector configuraiton to do load balancing. I'm not sure if re-running the wsconfig tool on all of the servers will do this or not, but that is what I would suggest trying first. If that doesn't work you will need to update the Tomcat connector configuraiton manually. You can find more information on load balancing with the Tomcat connector here: http://tomcat.apache.org/connectors-doc/generic_howto/loadbalancers.html.

ACE load-balancing-Cookie problem

In our other load-balancing environments the load-balancer-cookie contains the encrypted (real) servername or ip-address.
We think it's the same on the cisco, for that reason it's in theory not possible, that there are two 'green'-cookies with different values in the same request.
There are only two possibilities how this could happen:
a) The healthmonitor (http_probe) fails, the loadbalancer 'thinks' that the realserver is down and redistributes the traffic.
But in that case we would expect, that the old cookie will be overwritten by the new one and not simply added to the http-header.
b) The predictor in the serverfarm chooses a new realserver within the same request.
If that is really the cause of that problem this would be bug in the cisco ace.
What we found out, is that the loadbalancer performs a 'Set-Cookie'-Operation an every request even if the client submits the cookie correctly.
For example:
GET /ips-opdata/scripts/jquery.js HTTP/1.1
Host: www.xxxxx.com
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.9.2.15) Gecko/20110303 Ubuntu/10.04 (lucid) Firefox/3.6.15
Accept: */*
Accept-Language: en-us,en;q=0.5
Accept-Encoding: gzip,deflate
Accept-Charset: ISO-8859-1,utf-8;q=0.7,*;q=0.7
Keep-Alive: 115
Connection: keep-alive
Referer: http://www.xxxxx.com/
Cookie: green=R339366665; JSESSIONID=28D91FC6FD62A3921354BB36826294C4
HTTP/1.1 200 OK
Set-Cookie: green=R339366665; path=/; expires=Tue, 29-Mar-2011 06:33:00 GMT
Server: Apache-Coyote/1.1
X-Powered-By: Servlet 2.4; JBoss-4.2.2.GA (build: SVNTag=JBoss_4_2_2_GA date=200710221139)/Tomcat-5.5
ETag: W/"72181-1298537508000"
Last-Modified: Thu, 24 Feb 2011 08:51:48 GMT
Content-Type: text/javascript
Content-Length: 72181
Date: Mon, 28 Mar 2011 06:15:19 GMT
As you can see the cookies: green=R339366665 is transmitted from the client, but the loadbalancer does a Set-Cookie Operation of the same cookie once again. This is an unexpected behaviour.
We hope that this helps you to figure out the reason of the problem.

The cookie is sent by the ACE on each response to refresh the timeout value on the client. The value of the cookie doesn't change. This is the expected behaviour and shouldn't break anything in the application / browser.
For browser-based applications, don't forget to add the "browser-expire" parameter to your cookie-based stickyness config.

Load Balance Problem in Oracle RAC 10.2.0.4 (two nodes with CRS)

Hi all,
I have an Oracle RAC 10.2.0.4 with two nodes and I'm having problems with sessions load balance.
The problem is that all connections are going to the node which is enabled as 'master' in RAC, the node who has located .db resource in CRS.
If .db resource is located in node1 all connections are made in node1. In the same way, all connections are made in node2 when .db resource from CRS stack is located in node2.
The connections are made by a pool in Tomcat 6 using RAC service for made the connections.
Any idea?
Thanks in advance.
Here is an example of my net configuration:
listener.ora in node1:
LISTENER_RAC_NODE1 =
(DESCRIPTION_LIST =
(DESCRIPTION =
(ADDRESS = (PROTOCOL = TCP)(HOST = node1_vip)(PORT = 1525)(IP = FIRST))
(ADDRESS = (PROTOCOL = TCP)(HOST = node1)(PORT = 1525)(IP = FIRST))
tnsnames in node1:
LISTENERS_RAC =
(ADDRESS_LIST =
(ADDRESS = (PROTOCOL = TCP)(HOST = node1_vip)(PORT = 1525))
(ADDRESS = (PROTOCOL = TCP)(HOST = node2_vip)(PORT = 1525))
NODE1 =
(DESCRIPTION =
(ADDRESS = (PROTOCOL = TCP)(HOST = node1_vip)(PORT = 1525))
(ADDRESS = (PROTOCOL = TCP)(HOST = node1)(PORT = 1525))
(CONNECT_DATA =
(SERVER = DEDICATED)
(SERVICE_NAME = rac)
(INSTANCE_NAME = rac1)
RAC =
(DESCRIPTION =
(ADDRESS = (PROTOCOL = TCP)(HOST = node1_vip)(PORT = 1525))
(ADDRESS = (PROTOCOL = TCP)(HOST = node2_vip)(PORT = 1525))
(LOAD_BALANCE = yes)
(CONNECT_DATA =
(SERVER = DEDICATED)
(SERVICE_NAME = rac)
********************************************************

It is my fault.
It was the connection service. I had one one with preferred option and one without it. Now is working well.

How can I design Load Balancing for distant Datacenters? without single point of failure

Dear Experts,
We are using the following very old and passive method of redundancy for our cload SaaS but it's time to make it approperiate. Can youplease advise:
Current issues:
1. No load balancing. IP selection is based on primary and secondary IP configurations. If Primary fails to respond, IP record for DNS changes to secondary IP with TTL=1min
2. When primary server fails, it takes around 15 min for clients to access the servers. Way too long!
The target:
A. Activate a load balancing mechanism to utilized the stand-by server.
B. How can the solution be designed to avoid single point of failure? In the previous example, UltraDNS is a single point of failure.
C. If using GSS is the solution, how can it be designed in both server locations (for active redundancy) using ordinary DNS server?
D. How can HSRP, GSS, GSLB, and/or VIP be used? What would be the best solution?
Servers are running ORACLE DB, MS SQL, and tomcat with 2x SAN of 64TB each.

Hi Codlick,
the answer is, you cannot (switch to two web dispatchers).
If you want to use two web dispatchers, they need something in front, like a hardware load balancer. This would actually work, as WD know their sessions and sticky servers for those. But remember you always need a single point for the incoming address (ip).
Your problem really is about switchover groups. Both WD need to run in different switchover groups and need to switch to the same third software. I'm not sure if your switchover software can handle this (I'm not even sure if anyone can do this...), as this means the third WD needs to be in two switchover groups at the same time.
Hope this helps,
Regards,
Benny

JSF and load balancing issues

Hello,
We are having a difficult time getting our app to work correctly on our load balanced system. Any help / suggestions would be greatly appreciated.
We have a very basic load balancer which is in round robin mode (this cannot be changed), two separate vms, each with an instance of apache(2.0.54) and tomcat(6.0.14). Each apache talks only to the tomcat on the same vm and the tomcats are configured to session replicate with each other (which is happening).
When I log in to the app and start pressing a few buttons there are no problems for a little while. I then leave the screen idle for 1 minute and resume pressing buttons. The main session bean which holds information about where the user is and other key information, gets its constructor called and is set back to default and the page goes back to the main menu. Immediately before the bean is reset, the logs show the following:
[MyApp] 17 Jun 2009 15:33:27,514 WARN: [ajp-8009-2 LoggerListener.onApplicationEvent(60)] : Authentication event AuthenticationSuccessEvent: v; details: (etc etc removed…)This is what the two logs look like:
VM1 - Authentication event
VM2 - Main menu
VM2 - Presss a button – progress a screen
VM2 - Presss a button – progress a screen
VM2 - Presss a button – progress a screen
VM2 - Presss a button – progress a screen
[wait 1 min]
VM1 - Authentication event
VM1 - Bean reset
Processing continues, but screen is redirected to main menu
We have the distributable tag in the web.xml file and javax.faces.STATE_SAVING_METHOD is set to client. Its set to client as the app will not work at all with it set to server (keeps asking for the user to re-log in).
Thank you,
Victoria

Looks like a Tomcat issue to me. JSF is not going to timeout in 1 minute. You might want to check your Tomcat replication settings. Note try to do a very simple loadbalancing application that just echos the Tomcat is executing on.

Load Balancing Methods in Planning

Ware planning on Load Balancing our two web servers for performance reasons and our consultant and IT side had two different ideas of what this would mean.
Our IT side wants to us IIS and set up a Virtual Machine/IP that sits on both web servers (which are NIC card connected) thus creating the virtual entry point that would move activity to the least busy server.
Our consultant was thinking we'd have a seperate piece of HW (like CISCO or Baracuda - F5) that would perform the load balancing on it's own and direct traffic accordingly. He was not aware of an install utilizing the first method and we're wondering if this IIS solution will work or not.
We are using TOMCAT out of the box here too.
Any help is appreciated! We're toast on getting a new piece of HW in here in time for us to complete our migration but I just want to make sure we don't waste time going down a path that isn't viable.

Hi Paul,
The bulk of the work of a Planning server is performed by the Java App Server.
To have a highly available Planning environment you will need to look at clustering Tomcat and may need to use something like BEA Weblogic instead.
Check out the Clustering EPM Web Applications in the below PDF document.
http://download.oracle.com/docs/cd/E12825_01/epm.111/epm_manual_deployment.pdf
Regards,
-John

Sticky load balancing across 2 ports with cookies

Hi,
I have a server configuration where I have 1 top level Apache server that deals with SSL termination (and handles static content) and proxy passes dynamic content onto 2 Tomcat servers on 2 ports, one for http requests (9001) and one for the requests that were secure, but have now been un-encrypted by Apache (9002). My 2 Tomcat servers are load balanced using a CSS and I need this load balancing to stick to the tomcat servers regardless of port so that the user is stuck to the same Tomcat server for their entire session.
I would like to use arrowpoint cookies to perform this stickyness, but the documentation suggests that arrowpoint cookie load balancing (in fact any cookie based load balancing) requires the port to be specified in the content rule. Is this correct? Is my only option to use the source IP for stickyness? I don't understand why the port should be required if the stickyness is via a cookie. Can I not simply configure my 2 tomcat servers as services with no port and add a single content rule that load balances these services using arrowpoint-cookie advanced balancing?
service tomcat1
ip address x.x.x.x
active
service tomcat2
ip address x.x.x.x
active
owner me
   content sticky
     vip address x.x.x.x
     protocol tcp
     url "/*"
     add service tomcat-1
     add service tomcat-2
     advanced-balance arrowpoint-cookie
     active

Angela-
The issue with port is that cookies are very specifically HTTP only and the CSS has no way of knowing what protocol will hit a VIP prior to trying to address it as HTTP. Your issue is actually a bit clearer than it is initially led to be - you can still use 2 different rules by using the configuration below.
However, you might be headed for a headache if you don't implicitly control the client's actions. By default, browsers don't generally send cookies cross-protocol and definitely not cross-domain. Use something like httpwatch or iewatch to check out the headers your client sends to your site. Make sure when the 200ok arrives with the set-cookie that the client sends that cookie in all preceeding packets that are HTTP and HTTPS both.
service tomcat1
string "tomcat1"
ip address x.x.x.x
active
service tomcat2
string "tomcat2"
ip address x.x.x.x
active
owner me
   content sticky9001
     vip address x.x.x.x
     protocol tcp
     url "/*"
     port 9001
     add service tomcat-1
     add service tomcat-2
     advanced-balance arrowpoint-cookie
     active
   content sticky9002
     vip address x.x.x.x
     protocol tcp
     url "/*"
     port 9002
     add service tomcat-1
     add service tomcat-2
     advanced-balance arrowpoint-cookie
     active
With this configuration, the CSS will use the "string" as the cookie value. So if the client were to recieve set-cookie: ArrowpointCookie=tomcat1, it should use it for either rule, and end up on tomcat1 accessing either VIP.
Regards,
Chris

OC4J load balancing

We got multiple JVMs configured and running. Since we are not able to share the session among the JVMs we are having issue running the application on mutiple JVM since the OC4J is doing the load balancing. For example, if request is established on one JVM the next request is sent to another thus the application fails by not finding the session object. Is there a way we can have the sticky on OC4J so that if the session is established on one JVM then continue to use that JVM through out the life cycle of the session. Another word load balance only if the request is new. Thanks in advance for your help.

Okay, how about if you have multiple web application that are sharing the session.
Is there a way we could say these web application should be running on the same JVM? If you mean by this that you have different WAR files (either standalone or contained in an EAR file) that need to share common session state, then we don't support this with OC4J. A session is bounded within the Web module that it is created within. I think someone on here had come up with a way to do once before it using a different cookie name and a common cookie path but its not something we do by default.
Do you know how to configure them to run on the same JVM. Would parent work?
Could we make one web app parent for others - if so would it guarantee to run the the same JVM?The parent config doesn't dictate any runtime behaviour -- it's configuration, classloading based parenting only.
We have a scenario where our web applications are dependent upon each other - meaning they need to run on the same JVM together since they are sharing the session. It sounds to me that if you have this sort of requirement, then it'd be worth your while looking at our Coherence*Web functionality. This provides a way to allow the Coherence distributed cache to be used as the session/session-replication infrastructure -- effectively switching out the native session manager from OC4J (and other J2EE servers, Tomcat, etc.).
It has a switch that enables cross Web module state sharing. Because it's a distributed cache too, it will allow all JVMs to see/use the same view of the current session state.
See here for more details:
http://coherence.oracle.com/display/COH34UG/Coherence*Web+Session+Management+Module
http://coherence.oracle.com/display/COH34UG/Coherence*Web+Session+and+Session+Attribute+Scoping
-steve-

What is load balancing in sap business objects.

Hi friends.
Any one give me correct answer..

Hi,
it depends if you are talking about clustering the processing tiers (& using server groups) to achieve load-balancing, or if you mean using a load balancer in front of the web-tier (typically, Apache Tomcat) ? Also. it could be a software or hardware load balancer.
essentially, it splits the load across the boxes.
regards,
H

Load balancing Tomcat

Similar Messages

Maybe you are looking for