How to add files to a Fuse 6.1 profile using the CLI
For some mysterious reason its not possible to upload a file to a profile in Fuse 6.1. Since we tend to script the complete creation and deployment of our Fuse environments we needed to be able to add files to profiles from a script.
The support guys gave me this script that adds a function called ‘import-resource’, and since it really helped us I though I’d share it with the internet:
# Imports resources (ex. xml or properties) into a Fabric Profile # ex. # import-resource "TestProfile" "1.0" "something.xml" "/home/jsensharma/Downloads" # profile-display --display-resources TestProfile import-resource = { _profile = $1 _version = $2 _logical_file_path = $3 _physical_file_path = $4 _file_content = ((cat file://$_physical_file_path) | tac) _sref = $.context getServiceReference "io.fabric8.api.FabricService" _srv = $.context getService $_sref _dstore = $_srv getdatastore $_dstore setFileConfiguration $_version $_profile $_logical_file_path ($_file_content getbytes) }
After running this script you’ll be able to upload files like this:
import-resource test-profile 1.0 resource.data /tmp/resource.data
Running camel-tinkerforge on Karaf
Running the Camel-Tinkerforge component is really easy on Karaf. The ready to go feature files make installation easy, and the use of the auto-deployment directory makes testing different route configurations also easy.
The camel-tinkerforge component is part of the Camel-labs IoT project. We just got started with this so the first release is still pending. This means you need to checkout the code from https://github.com/camel-labs/camel-labs and run “mvn install -DskipTests=true”. There’s no need to run the tests since the buildserver has already done that. 🙂
To intall Camel-Tinkerforge on Karaf 2.x run the following commands in the Shell:
features:chooseurl camel 2.15.2 features:install camel-blueprint features:addurl mvn:com.github.camel-labs/camel-tinkerforge/0.0.0-SNAPSHOT/xml/features features:install camel-tinkerforge
Now create a file with the following content:
<blueprint xmlns="http://www.osgi.org/xmlns/blueprint/v1.0.0"> <camelContext xmlns="http://camel.apache.org/schema/blueprint"> <route> <from uri="tinkerforge:/temperature/T9"/> <to uri="log:default"/> </route> </camelContext> </blueprint>
Run log:tail in the karaf Shell and copy the xml file to the deploy directory in the karaf installation directory. Notice how Karaf starts the camel routes defined in the xml file, and writes messages to the log whenever the temperature changes.
Now you can start experimenting with different routes in the xml file, and as soon as you save the file Karaf will automatically reload the routes.
Remote Debug on Fuse Fabric Container
When you create a container in fabric you can specify which options should be passed to the container. This way you can set the amount of memory for instance, but you can also specify that the container should start in debug modus.
The following Fuse CLI command would create a container with a maximum heap size of 256 MB:
container-create-child --jvm-opts "-Xmx256m" root child
In the Hawt.io console you’ll find the “Jvm opts” option under the “Advanced” tab on the container creation page.
Now you can also use this to pass the regular debug commands to a JVM: “-Xdebug -Xrunjdwp:transport=dt_socket,address=5001,server=y,suspend=n”.
The CLI command would then look like this:
container-create-child --jvm-opts "-Xdebug -Xrunjdwp:transport=dt_socket,address=5001,server=y,suspend=n" root child
The suspend=n part means that the JVM won’t wait for a debugger to connect but will startup like usual. If you need to see what happens in the JVM right after it starts you can change this to suspend=y. The JVM will then block until a debugger is connected.
This post is based on my answer to luca.santaniello.81 on the JBoss Developer forums.
How to control the fan speed on the Odroid U3
I noticed the (optional) fan on my Odroid U3 frequently spinning up only to shut down again 1 second later. This appeared rather buggy to me so I looked into how to control the fan. This is what I found:
You can switch the fan into ‘auto’ or ‘manual’ mode:
echo auto > /sys/devices/platform/odroidu2-fan/fan_mode
echo manual > /sys/devices/platform/odroidu2-fan/fan_mode
Note: The ‘odroidu2’ part of the path is not a typo! Strangely enough the U3 uses the same device path as the U2.
Once you’ve switched the fan into manual mode, you can set the speed of the fan by echo-ing a value between 0 (0%) to 255 (100%) to the fan controller:
echo 0 > /sys/devices/platform/odroidu2-fan/pwm_duty (fan off)
echo 255 > /sys/devices/platform/odroidu2-fan/pwm_duty (fan on full speed)
Fun fact: You can get into an argument with the automatic fan controller by setting the fan speed while in auto mode. The fan will respond to your command but the automatic controller will shut it down after noticing what you did. 🙂
While playing with the fan use the following command to read the CPU temperature:
cat /sys/devices/virtual/thermal/thermal_zone0/temp
And to run a CPU stress-test that utilizes all 4 cores use this command:
sudo openssl speed -multi 4
User xana (among others) has posted a script on the odroid u3 forum that uses the commands described above to periodically set the fan speed based on CPU temperature: http://forum.odroid.com/viewtopic.php?f=83&t=6275
Update:
If you have the latest version of the kernel it gets even simpler. You can set the temperature above which the fan should be used to cool down the SoC:
echo 60 > /sys/bus/platform/devices/odroidu2-fan/start_temp
How to easily switch the default Java version on Mac OSX
After installing Netbeans and Java7 for some JavaFX tinkering several of my other apps stopped working.
A lot of the solutions I found online involve writing scripts or changing startup scripts for specific programs, but I found none of them easy or maintainable enough. I ended up altering the symbolic link named “Current” in /System/Library/Frameworks/JavaVM.framework/Versions
A link called CurrentJDK in the same directory still pointed at a 1.6 version, so I simply changed the Current link to point to the CurrentJDK link. That way I can simply change it back when I want 1.7 to be the default.
sudo ln -s CurrentJDK/ Current
How to install Glassfish 3.1 Clustering on multiple hosts using linux command line
Today a colleague needed help with setting up a Glassfish cluster and I remembered having read a very good tutorial that created the entire cluster using only asadmin commands. In my experience using only asadmin commands not only gives you a greater understanding of the way Glassfish works internally but also enables you to create scripts for Glassfish administration commands.
Anyway, I managed to find the blogpost that I read about a year ago and decided that it was a good link to share with the world: http://www.randombugs.com/java/glassfish/installing-glassfish-31-cluster-debian-60-command-line.html
How to connect Glassfish 3 to an external ActiveMQ 5 broker
Introduction
Here at ONVZ we’re using Glassfish 3 as our development and production application server, and we’re quite happy with its performance and stability, as well as the large community surrounding it. I rarely run into a problem that does not have a matching solution on stackoverflow or java.net. As part of our open source strategy we also run a customized ActiveMQ cluster called “ONVZ Message Bus”.
To enable Message Driven Beans and other EJBs to consume and produce messages to and from the ActiveMQ message brokers, ignoring the internal OpenMQ broker that comes shipped with Glassfish, an ActiveMQ Resource Adapter has to be installed. Luckily for me Sven Hafner wrote a blog post about running an embedded ActiveMQ 5 broker in Glassfish 3, and I was able to distill the information I needed to connect to an external broker instead. This blog post describes what I did to get it to work.
Maven best practices
Although Maven offers a “convention over configuration” solution, there is still more then enough configuration necessary to cause some serious headache. In this post I will share some best practices with you that will ease maintenance of your POM files.
How to post nice looking source code on WordPress.com
If you want to post source code and want to make it easy for your audience to read it, use Syntax Highlighter. Blogs on WordPress.com can use this without having to install anything. See http://en.support.wordpress.com/code/posting-source-code for more information.