Continues Integration for Hazelcast

I have set up a Hudson as a CI tool for Hazelcast. It is hosted on EC2.  

Whenever there is a new commit, it will 

• compile hazelcast
• run tests
• generate Emma code coverage trends
• generate FindBugs trend.

All anonymous users are able to see the build history. 

I found Hudson very easy to setup. It integrates very well with maven. 

Happily it has lots of plugins. I have installed couple of them, but would appreciate if you suggest any that you think we should have. 

Here is the link: http://184.73.226.176/hudson/

Hazelcast Cluster Monitoring Tool

Last week we were busy with new release, Hazelcast 1.8.1. Finally we made it and shipped the most robust, feature-rich and still fastest Hazelcast ever. This release comes with Cluster Monitoring Tool.

Cluster Monitoring Tool is a simple war file that you can deploy on any web container and start monitoring your running cluster. See documentation for more details. People ask whether it is open source or not. Of course it is. We want you to look at the code, to report us bugs, suggest enhancements and share your opinion. This is what makes Hazelcast very valuable.

As your cluster gets bigger and bigger, it becomes harder to keep up with logs and see how it performs. You know that your cluster is running somewhere there. But you can not see and feel it. The aim of Monitoring Tool is to make you see the cluster. It aggregates all relevant information from all members and shows you in a simpler way.  

We were dreaming of it for a long time and now we have it thanks to Native Java Client. Monitoring tool uses Java client to connect to the cluster and periodically to get some information to display.

We are very amazed with it's abilities. It shows the internals of the running cluster. When it first showed up, we were able not only find some bugs, but see the bugs. It helped a lot to change some logic that we didn't like.

Here is couple of things you can use it for. 


From the tool you can see how your data is distributed, how many operations per second each node is performing. If the key's are string, you can even make a get operation, using Map Browser, and see details of that entry.

But those of you using map.lock() can watch the total number of locked entries on each node using the Size&Memory Details Panel. It also shows total number of threads, waiting for those locks. These threads are your user threads, so you can see how much your application is waiting for locks. , You may consider to change it if there is lots of threads waiting.

Here is the screenshot from it:

Now it is time to try it out. We are looking for your feedback.