Jump to: navigation, search

Difference between revisions of "Stardust/Knowledge Base/Getting Started/RTE on AWS"

(Process Model Creation and Deployment)
(Set Up)
 
(27 intermediate revisions by the same user not shown)
Line 16: Line 16:
 
[[File:InstanceTypes.gif]]
 
[[File:InstanceTypes.gif]]
  
The AWS Virtual Machine Image (AMI) is configured to run on all Instance Types including the '''Standard Micro (t1.micro)''' type and your choice depends entirely on the expected load profile. For first experiments the '''Standard Micro''' is definitely sufficient.
+
The AWS Virtual Machine Image (AMI) is configured to run on all Instance Types including and above the '''Standard Small (m1.small)''' type and your choice depends entirely on the expected load profile. For first experiments the '''Standard Small''' is sufficient, however with increased use, you will soon see issues because of the limited memory (see '''VM Settings''').
 +
 
 +
Note, that the image would not run stable on a '''t1.micro''' type, just because of its memory limitations.
  
 
Please consider that although there are no charges for Stardust, '''Amazon charges you for most of the EC2 configurations'''.
 
Please consider that although there are no charges for Stardust, '''Amazon charges you for most of the EC2 configurations'''.
Line 28: Line 30:
 
[[File:KeyPair.gif]]
 
[[File:KeyPair.gif]]
  
You only need this if you intend to login to the EC2 instance and modify the setup, which is needed e.g. for
+
You need a Key Pair this if you intend to login to operating system of the EC2 instance via Secure Shell (SSH) and modify the setup there, e.g. for
  
 
* troubleshooting (''see below''),
 
* troubleshooting (''see below''),
Line 36: Line 38:
 
Hence, we recommend to generate Key Pair and [http://docs.aws.amazon.com/gettingstarted/latest/wah-linux/getting-started-deploy-app-connect.html and use it].
 
Hence, we recommend to generate Key Pair and [http://docs.aws.amazon.com/gettingstarted/latest/wah-linux/getting-started-deploy-app-connect.html and use it].
  
If you just intend to run and configure Stardust via the Stardust Web Portal, there is no need to create a Key Pair and you can consider the installation a block box.
+
If you just intend to run and configure Stardust via the Stardust Web Portal, there is no need to create a Key Pair and you can consider the installation a black box.
  
 
=Login=
 
=Login=
Line 51: Line 53:
 
This is the Stardust root user. Make sure that you can [http://help.eclipse.org/kepler/topic/org.eclipse.stardust.docs.enduser/html/handbooks/execution-clients/admin-portal/views/admin-portal-participant.html?cp=52_9_0_4_12 change the password later].
 
This is the Stardust root user. Make sure that you can [http://help.eclipse.org/kepler/topic/org.eclipse.stardust.docs.enduser/html/handbooks/execution-clients/admin-portal/views/admin-portal-participant.html?cp=52_9_0_4_12 change the password later].
  
Use the Stardust Portal as described in the respective [http://help.eclipse.org/kepler/topic/org.eclipse.stardust.docs.enduser/html/handbooks/execution-clients/carnot-clients.html?cp=52_9_0 chapter] of the Stardust documentation.
+
After successful login you can [http://help.eclipse.org/kepler/topic/org.eclipse.stardust.docs.enduser/html/handbooks/execution-clients/carnot-clients.html?cp=52_9_0 use the Stardust Portal].
  
 
=Process Model Creation and Deployment=
 
=Process Model Creation and Deployment=
  
If you are keen to start playing and do not bother to create a Process Model at this point pick one of the [http://wiki.eclipse.org/Stardust/Knowledge_Base/Sample_Models sample process models] from Git e.g. the [http://git.eclipse.org/c/stardust/org.eclipse.stardust.examples.git/plain/tutorials/modeling/CarRental.xpdl Car Rental Example] (right-click the link before and save the file) and just upload and deploy under '''Administration > Model Management''' (more details [http://help.eclipse.org/kepler/topic/org.eclipse.stardust.docs.enduser/html/handbooks/execution-clients/admin-portal/views/admin-modelmanagement.html?cp=52_9_0_4_8 here]).
+
If you are keen to start playing and do not bother to create your own Process Model at this point, pick one of the [http://wiki.eclipse.org/Stardust/Knowledge_Base/Sample_Models sample process models] from Git e.g. the [http://git.eclipse.org/c/stardust/org.eclipse.stardust.examples.git/plain/tutorials/modeling/CarRental.xpdl Car Rental Example] (right-click the link before and save the file) and just upload and deploy under '''Administration > Model Management''' (more details [http://help.eclipse.org/kepler/topic/org.eclipse.stardust.docs.enduser/html/handbooks/execution-clients/admin-portal/views/admin-modelmanagement.html?cp=52_9_0_4_8 here]).
  
 
[[File:Model-deployment.gif]]
 
[[File:Model-deployment.gif]]
Line 64: Line 66:
 
Do not forget to [http://help.eclipse.org/kepler/topic/org.eclipse.stardust.docs.enduser/html/handbooks/execution-clients/admin-portal/views/admin-portal-participant.html?cp=52_9_0_4_12 create users and assign to roles] so that you can start processes and perform activities in your process.
 
Do not forget to [http://help.eclipse.org/kepler/topic/org.eclipse.stardust.docs.enduser/html/handbooks/execution-clients/admin-portal/views/admin-portal-participant.html?cp=52_9_0_4_12 create users and assign to roles] so that you can start processes and perform activities in your process.
  
=Troubleshooting=
+
=VM Settings=
Below a few things which might help you if you are facing problems. If nothing helps, post in the [http://www.eclipse.org/forums/index.php?t=thread&frm_id=225 Stardust Forum]. You very likely find help there.
+
==Tomcat Out of Memory Issue==
+
Currently, the Tomcat installation on the Stardust Image is not configured with the appropriate memory size for its Java VM. To address this issue, [http://docs.aws.amazon.com/gettingstarted/latest/wah-linux/getting-started-deploy-app-connect.html login to your EC2 Instance using your Key Pair] and add the following
+
  
<source lang="bash">
+
The Tomcat installation of the Stardust EC2 Image is configured with a minimal memory size for its Java VM which fits with the settings of a '''Standard Small (m1.small)''' Instance Type. For broader use (more users, high process volumes) you may need to adjust the settings.
set CATALINA_OPTS=%CATALINA_OPTS% -Dtrue=true -Xms1200m -Xmx1200m -XX:NewSize=300m -XX:MaxNewSize=300m -XX:PermSize=256m -XX:+UseTLAB
+
</source>
+
  
at the beginning of '''/home/ec2-user/apache-tomcat-6.0.32/bin/catalina.sh'''.
+
Insufficient memory configuration will show as sudden unavailability of the application in the browser which can only be recovered via reboot.
 +
 
 +
[http://docs.aws.amazon.com/gettingstarted/latest/wah-linux/getting-started-deploy-app-connect.html Login to your EC2 Instance using your Key Pair] and modify/add the following to '''/home/ec2-user/apache-tomcat-6.0.32/bin/catalina.sh''' depending on the expected load:
 +
 
 +
{| border="1" cellspacing="0" cellpadding="5"
 +
|-
 +
! Number of parallel Users
 +
! Machine Configuration (Matching EC2 Instance)
 +
! catalina.sh Settings
 +
|-
 +
| 5
 +
| 1 CPU, 1.7 GB RAM '''(m1.small)'''
 +
| <source lang="bash">set CATALINA_OPTS=%CATALINA_OPTS% -Dtrue=true -Xms800m -Xmx800m -XX:NewSize=200m -XX:MaxNewSize=200m -XX:PermSize=256m -XX:+UseTLAB -XX:+UseParNewGC -XX:+UseConcMarkSweepGC</source>
 +
''(Default Settings)''
 +
|-
 +
| 50
 +
| 4 CPU 6-8 GB RAM '''(m1.large)'''
 +
| <source lang="bash">set CATALINA_OPTS=%CATALINA_OPTS% -Dtrue=true -Xms4g –Xmx4g -XX:NewSize=1g -XX:MaxNewSize=1g -XX:PermSize=256m -XX:+UseTLAB -XX:+UseParNewGC -XX:+UseConcMarkSweepGC</source>
 +
|-
 +
| 100
 +
| 8 CPU 12 GB RAM '''(m1.xlarge)'''
 +
| <source lang="bash">set CATALINA_OPTS=%CATALINA_OPTS% -Dtrue=true -Xms8g –Xmx8g -XX:NewSize=2g -XX:MaxNewSize=2g -XX:PermSize=256m -XX:+UseTLAB -XX:+UseParNewGC -XX:+UseConcMarkSweepGC</source>
 +
|-
 +
| > 100
 +
| >= 8 CPU >= 12 GB RAM
 +
| ''(see above)''
 +
|}
 +
 
 +
Consider the [http://help.eclipse.org/kepler/topic/org.eclipse.stardust.docs.dev/html/handbooks/operation/tuning/ag-tuning-1.htm?cp=52_6_3_2 Stardust tuning recommendations] for details.
 +
 
 +
=Troubleshooting=
  
''This problem will be addressed soon on the original image.''
+
Below a few things which might help you if you are facing problems. If nothing helps, search and/or post in the [http://www.eclipse.org/forums/index.php?t=thread&frm_id=225 Stardust Forum]. You will very likely find help there.
  
==Other Trouble Shooting==
+
==Brute Force Troubleshooting==
 
If something totally unexpected happens in the Portal, try a reboot of your EC2 instance after making sure that no other user is working on the system.
 
If something totally unexpected happens in the Portal, try a reboot of your EC2 instance after making sure that no other user is working on the system.

Latest revision as of 10:03, 17 October 2013

Set Up

To create a full Stardust Runtime Environment on Amazon Web Services EC2 Infrastructure,

ProductPage.gif

You will be asked to login with an existing Amazon account or to register for one:

AmazonLogin.gif

Launch with one 1-Click will launch the Stardust Runtime on your EC2 Instance Type of choice:

InstanceTypes.gif

The AWS Virtual Machine Image (AMI) is configured to run on all Instance Types including and above the Standard Small (m1.small) type and your choice depends entirely on the expected load profile. For first experiments the Standard Small is sufficient, however with increased use, you will soon see issues because of the limited memory (see VM Settings).

Note, that the image would not run stable on a t1.micro type, just because of its memory limitations.

Please consider that although there are no charges for Stardust, Amazon charges you for most of the EC2 configurations.

Select your data center region of choice - the Stardust AMI is available for all AWS data center regions:

Region.gif

You will be prompted to confirm that you do not want to create a Key Pair:

KeyPair.gif

You need a Key Pair this if you intend to login to operating system of the EC2 instance via Secure Shell (SSH) and modify the setup there, e.g. for

  • troubleshooting (see below),
  • tuning for larger deployments or
  • adding application specific Java byte code.

Hence, we recommend to generate Key Pair and and use it.

If you just intend to run and configure Stardust via the Stardust Web Portal, there is no need to create a Key Pair and you can consider the installation a black box.

Login

After you have launched your instance and the instance is available, lookup the Public DNS Name for the instance and login via

http://<Public DNS Name>:8080/stardust-standard

and

  • User: motu
  • Password: motu

This is the Stardust root user. Make sure that you can change the password later.

After successful login you can use the Stardust Portal.

Process Model Creation and Deployment

If you are keen to start playing and do not bother to create your own Process Model at this point, pick one of the sample process models from Git e.g. the Car Rental Example (right-click the link before and save the file) and just upload and deploy under Administration > Model Management (more details here).

Model-deployment.gif

If you intend to create your own Process Models, you can either use the Browser Modeler which is immediately available in the Stardust Portal of your Amazon EC2 Instance or use the Stardust Eclipse Modeler - which requires you to download Eclipse and install Stardust.

User Creation and Role Assignment

Do not forget to create users and assign to roles so that you can start processes and perform activities in your process.

VM Settings

The Tomcat installation of the Stardust EC2 Image is configured with a minimal memory size for its Java VM which fits with the settings of a Standard Small (m1.small) Instance Type. For broader use (more users, high process volumes) you may need to adjust the settings.

Insufficient memory configuration will show as sudden unavailability of the application in the browser which can only be recovered via reboot.

Login to your EC2 Instance using your Key Pair and modify/add the following to /home/ec2-user/apache-tomcat-6.0.32/bin/catalina.sh depending on the expected load:

Number of parallel Users Machine Configuration (Matching EC2 Instance) catalina.sh Settings
5 1 CPU, 1.7 GB RAM (m1.small)
set CATALINA_OPTS=%CATALINA_OPTS% -Dtrue=true -Xms800m -Xmx800m -XX:NewSize=200m -XX:MaxNewSize=200m -XX:PermSize=256m -XX:+UseTLAB -XX:+UseParNewGC -XX:+UseConcMarkSweepGC

(Default Settings)

50 4 CPU 6-8 GB RAM (m1.large)
set CATALINA_OPTS=%CATALINA_OPTS% -Dtrue=true -Xms4g –Xmx4g -XX:NewSize=1g -XX:MaxNewSize=1g -XX:PermSize=256m -XX:+UseTLAB -XX:+UseParNewGC -XX:+UseConcMarkSweepGC
100 8 CPU 12 GB RAM (m1.xlarge)
set CATALINA_OPTS=%CATALINA_OPTS% -Dtrue=true -Xms8g –Xmx8g -XX:NewSize=2g -XX:MaxNewSize=2g -XX:PermSize=256m -XX:+UseTLAB -XX:+UseParNewGC -XX:+UseConcMarkSweepGC
> 100 >= 8 CPU >= 12 GB RAM (see above)

Consider the Stardust tuning recommendations for details.

Troubleshooting

Below a few things which might help you if you are facing problems. If nothing helps, search and/or post in the Stardust Forum. You will very likely find help there.

Brute Force Troubleshooting

If something totally unexpected happens in the Portal, try a reboot of your EC2 instance after making sure that no other user is working on the system.