systec.blogsite.org

From November 10 to November 11 we have joined the “EntwicklerTage 2008″ in Munich as I had posted a few time before. It was a really great workshop with a lot of insider information from the Apache Tomcat Project. We had a lot of fun with Peter who made the workshop a funny unique event!

He started with a basic overview of the Tomcat development process. After that we learned about the internal work of the Tomcat including the newest features like http executors. Next to this we build up a Tomcat configuration an go through a lot of options.

At the workshop we were 10 people. All colleagues we met where really cool guys and we had the chance to talk a lot about our work and our different experiences and problems.

I recommend this workshop to everyone who is working with Tomcat in a productive environment but be aware! Yes, it was a administrator workshop but you have to bring basic programming knowledge along with you!

Börnie made some photos with his digicam. I think he will append it here.

… but there was a lot of work to manage. In the last month we had to get a lot of things off the ground, and I hope that I will find time do document it here.

Some buzzword topics from the last month:

1. OpenVZ news
2. Oracle RMAN Online Backup
3. Oracle Database Updates
4. Tomcat Workshop with Peter Rossbach
5. Tomcat Template System
6. different Web Projects
7. PCVisit and Tunneling
8. …

Finaly, the site is still alive! Please be patient with us

Regards…

As you eventually know, we are using about 70 Apache Tomcats in our web server farm. No we are facing new challenges in this farm. In the last week we got some requests to build up some virtual private serves (VPS) with up to 100 GB (gigabytes) disk space.

Our farm is completely virtualisized. For virtualization we are using the open source basis from Parallels Virtuozzo Containers called OpenVZ.

We have no problem to provide terrabytes of storage space because we have to EVA 6000 storages but I would like to use this disk space as real shared storage. I think I will try it with lustre cluster file system.

But why not with GFS2 or OCFS2? Because GFS2 kernel patches could not integrated with the OpenVZ patches (see OpenVZ forums) and OCFS2 doesn’t provide disk quotas.

It seems like that lustre is the only working combination with OpenVZ kernel patches.

Maybe it will work – maybe it won’t – we will see.

I would also like to use FAI for faster deployment of the OpenVZ hosts based on Debian.

If I will find time I will document the results here. If anyone read this post and had already tried this setup, please let me know!

On Wednesday 10 this week we had our Skype meeting with Peter Roßbach. Peter Roßbach is commiter from the Apache Tomcat project. In our company we have a large tomcat farm with about 80 tomcats which are powering different applications.

For load balancing we use pound at the moment but we would like to migrate to Apache AJP1.3 as load balancer. Apache is a more flexible way because we will have the availability change the running configuration without loosing the actual connections.
Three years ago we tried Apache as load balancer but we had no success with it. But, I have to say that we hadn’t enough experience with it.

So my idea was to contact Peter Roßbach. I fist met him at the Linux Day 2008 in Berlin. He had a conference about Apache high availability and clustering. Now I contact him and I must say that he is a very impressive guy. He was the first one who was able to follow me through different regions in a few phrases.
By now I thought that I can configure Apache Tomcats and AJP 1.3. but he showed my limits.

I hope that I will have the chance to learn a lot from Peter because I think he one of the few who knows nearly everything of Apache and Tomcat.

Michael Prokop from the grml project sad in a conference at the Grazer Linux Tag 2008: “Always try to learn from the best” – Yes I will try!

If could I will post the resulting configuration an techniques here.