Tuesday Dec 14, 2010

To ssh or not to ssh? (with GlassFish 3.1)


With GlassFish 3.1, a good deal of the centralized administration (one of the corner-stones of clustering) seem to depend on the use of SSH. In a recent blog post, Joe Di Pol discusses the benefits of relying on SSH but also ways to work around it if your environment doesn't offer easy SSH configuration.

The blog is really a technical article that walks you through the basics of SSH if you're not fully familiar with it yet, and how to set it up (including the authentication part) on environments such as windows where such things don't come as easy as on other operating systems.

One of the neat things ssh enables is remote installation of GlassFish instances (asadmin install-node) and nodes (asadmin create-node-ssh). It also covers how you can do without SSH and yet still be able to talk to remote instances. So be sure to bookmark Joe's post if you're at all interested in GlassFish clustering.

Friday Nov 26, 2010

It's raining GlassFish 3.1 Cluster screencasts!

A couple of months back (soon after Milestone 3 was released), we published a "GlassFish 3.1 Admin Console" screencast which offered a glimpse at what centralized admin and clustering would look like in the GlassFish 3.x family. As recently as last week, another GlassFish HA video was published based on the more recent bits.

Earlier this week, Paulo released a detailed three-part screencast series in Brazilian Portuguese : [1], [2] and [3] based on these full instructions and instances running in virtualbox images and accessed via ssh. And finally we now have Sébastien publishing a short (7-min) demo with GlassFish 3.1, a load-balancer and session failover from one node to another (and back) in 12-steps. Clearly Sébastien decided to have a bit of fun in the post-processing of his video. Check it out!

Maybe we should start a competition for the best GlassFish 3.1 screencast and award prizes when the product ships? The GlassFish Youtube channel has more than 325 subscribers with close to 15K channel views and 60K upload views.  Subscribe to not miss any of the content; so far there are 60 videos.

Saturday Nov 20, 2010

From Download to Testing a Clustered Deployment in Under 10 Minutes

Arun's latest screencast shows off GlassFish 3.1's HA and Centralized Administration.

The video shows how to create a 2-instance cluster on a single machine using the web-based Administration Console, deploy the canonical clusterjsp application, and show session fail-over capabilities in GlassFish 3.1.  From download to testing an application with high availability all in under 10 minutes...

Arun's video is at our official YouTube channel; follow the links to GlassFishVideos, or directly to GlassFish 3.1 Clustering, High Availability and Centralized Administration.

As of the writing this note, the channel has more than 300 subscribers with more than 14K channel views and more than 56K upload views.  Subscribe to not miss any of the content; so far there are 51 videos.

Friday Oct 22, 2010

A practical guide to configuring and testing GlassFish 3.1 Clustering

GlassFish Clustering

The main theme for GlassFish 3.1 is clustering which really encompasses centralized admin, load-balancing and in-memory state replication (HA). These features are all available in the 2.x family and are now being introduced in the OSGi-based and JavaEE6-compatible GlassFish product.

While a lot of engineering time has been spent on making the clustering configuration as easy as possible (even with v2.x, see this short demo), one will need to test the configuration and adapt to the networking environment so that's why Shreedhar wrote this recent entry on "How To Configure & Test High Availability with GlassFish Server 3.1 Using A Single Machine"

This blog entry discusses the following :
• the new asadmin validate-multicast command
• creation of a glassfish cluster with explicit multicast address and port
• use of the asadmin get-health command to ensure an HA cluster was properly started
• deployment of applications with --availabilityenabled=true and testing the fail-over
• differences in choosing the replica instances compared to GlassFish v2.x behavior
• how file issues and provide feedback

What may not appear in the list above is the amount of work that went into developing and enhancing the runtime clustering through Shoal's GMS and In-memory replication features (check the one-pagers for more details: In-memory Session Replication, GMS Runtime Clustering, GMS Configuration).

With Milestone 6 scheduled for early next week and the FishCAT program just kicking off (thanks to Richard and other community members!), now is a great time for you to start testing your applications in a GlassFish 3.1 HA environment.

Monday Jun 28, 2010

More GlassFish 3.1 Progress - M2 and Brown Bags

Summer is here (and so is the World Cup), but the traffic on the GlassFish aliases is going up as we continue to move towards GlassFish 3.1.

Last week was Milestone 2 which included several demos: SSH-based cluster management, Deployment and WebSockets (scripts only so far, screencasts on the way).  See Carla's post on creating and starting instances on remote hosts and also Arun's detailed TOTDs based on M2: #141: Running on Ubuntu AMI on Amazon EC2  and #142: SSH Provisioning and Start/Stop Instance/Clusters.

Also, we have restarted the Brown Bags for the development team (we did a few back in 2007). Jerome gave the first one (on Config) 3 weeks ago, and tomorrow Tom, Bhakti and Byron will present on Cluster Infrastructure Dev Test development. The BBags are intended for the DEV team but are open to all; check Rajiv's announcement for details.

Saturday May 29, 2010

GlassFish 3.1 M1 Available: Binaries, Demos and Screencasts

Milestone 1 of the GlassFish 3.1 Open Source Edition is now available from our promoted page as 3.1-b02: full distro (unix, windows, zip) and web distro (unix, windows, zip).

GlassFish has been following a Milestone model for many years and it works very well; for 3.1 we are documenting better each milestone (M1 content) and adding two new twists: demos to explain the new features, and community updates to enable Q&A.

Two of the features of M1 are Basic Clustering with Dynamic Reconfiguration, and Application Versioning.  Each comes with its own demo which Tom (Clustering and Reconfiguration) and Hong (Versioning) presented yesterday to a small team (also see Arun's commentary). The demos are quite easy to set up; you can replicate them yourselves or you can go directly to Alexis's screencasts.

The Clustering Demo  is very well documented; see Wiki page and associated WAR.  Also see Arun's commentary as well as Alexis' screencast at YouTube (5:38').

I need to chase down the actual demo used for the Versioning piece - it was neat in that each version of the Web App deployed was a simple Hello World using different technologies: Servlet, JSF, JSP, EJBs, POJOs... The funtionality itself is very well documented: Basic Commands and Design Spec, and general page on Collaboration with Serli. For the demo, see Arun's commentary as well as Alexis' screencast at YouTube (5:02').

We will cover M1 and both screencasts at our next community update.  Tentative date is 9am PT, Tuesday, June 8th.

Saturday Aug 29, 2009

New Webinar Recording: Load Balancing GlassFish Clusters


Last month, John and Shreedhar gave a free webinar on High Availability through Load Balancing of GlassFish Clusters. The presentation was well attended, see posts from John and from an attendee, Daniel-Noz, and the recording of the webinar is now available.

As part of an effort to improve the effectiveness of this and other GF Portfolio Resources, I've created a very short Feedback Survey; please consider using it to help us help you.

And, if you are interested in this topic, also check out this White Paper Application Scalability and Fault Tolerance with GlassFish Load Balancer (and companion survey).

Wednesday Apr 29, 2009

WebSpace Server Monitoring and Performance

WebSpace Server Homepage

Rajesh and team are hard at work plugging in more JMX monitoring features in WebSpace Server and Liferay (see the detailed proposal). This will give administrators a nice overview of performance metrics, fault diagnostics, and in general keeping WebSpace running smoothly. Speaking of performance, Srikanth has also written up an integration of WebSpace Server with Apache Solr, which is recommended for use with GlassFish clusters. The WebSpace Wiki is a great source of material like this for developers and administrators.

Saturday Dec 06, 2008

John Shows GlassFish v2 Clustering in 10 Minutes


John just posted a very nice screencast showing how to go From Zero to a v2 Cluster in 10 minutes, including download, asadmin configuration, GUI configuration, and tests. Pretty neat, check it out.

Easy configuration of Clustering is one of the assets of GlassFish, and, as users move into production, the interest is growing. A couple of weeks ago I highlighted some Posts and Slides on Clustering and also uploaded slides to SLX and SlideShare and they have proven quite popular.

Given the interest, I have started collecting links for a Proposed Webinar Set on the topic; hopefully in January.

Saturday Nov 22, 2008

Clustering in GlassFish v2 and v2.1


Clustering - supported in GlassFish since Sept 2007 - will be further improved in the GFv2.1 release (companion to SailFin). One of the improvements is Apache 2.2 support and Kshitiz has written an good writeup on the configuration steps and mentions the automated support also.

Clustering is used for horizontal scalability and high availability; interest and use of the feature is increasing quickly with GlassFish deployments. Check the official documentation and our recent White Paper on HA. Also, Satyajit just put together a set of slides on the topic (writeup, PDF@SLX, SlideShare).

We will cover Clustering in a future TA Online event; the Seam Online Event on Thursday went very well and I think the new format with multiple, shorter, presentations around a single topic is well suited to this new medium.

Tuesday Aug 26, 2008

Shoal 1.1 update available

Shoal of Fish

Shreedhar is announcing a new build release for Shoal 1.1, the GlassFish dynamic clustering framework. Version 1.1 is work in progress and offers a JoinedAndReadyNotificationSignal, multiple clusters support for the same member (useful for Sailfin's Converged Load-Balancer), cross subnet clustering support, etc.

The most important driver for the Shoal project lately has been the Sailfin project (now in alpha stage). Compared to HTTP, SIP sessions are smaller in size but greater in number so this has been a very interesting set of constraints for this group management and data replication technology.

If you're new to the technology, Shoal is at the heart of the GlassFish v2 clustering architecture and uses JXTA as an underlying technology. This article is a nice introduction if you're new to the technology which is also used in other projects such as GreenFire.

Tuesday Apr 29, 2008

GlassFish cluster administration - When do I talk to the DAS?

Small schema of cluster

Kedar has a blog on GlassFish cluster administration (domain, DAS, Node Agents, Instances, ...). This should come handy to people looking at understanding when synchronisation actually happen with the DAS (Domain Administration Server) which is where the centralized repository sits.

We often claim to have BEA's features at JBoss' price. This centralized cluster management is certainly one such feature.

Sunday Dec 09, 2007

New Shoal Article on java.net

Project Shoal

Most GlassFish subprojects (Grizzly, Metro, Woodstock, HK2, Jersey, mq, Mojarra, ...) can be used independently from the application server. Project Shoal is no exception. This Group Management Service used by GlassFish clustering can also be applied to other use-cases.

Community members Juan Pedro Danculovic and Diego Naya have a nice article over on java.net covering in a single place most of the Shoal features. It goes into the definition of all the basic terms such as Group, Member, Member Token, Spectator, Core, Group Management Service, Components, and Signals. The article also covers how to integrate your own code into the clustering infrastructure (not only for Java EE applications, but SE applications too). Given the artifact's size (JARs are less than 2MB), together with a stripped-down GlassFish v3, the possibilities become very interesting.

Previous Shoal resources include:
•  Previous Shoal posts
•  Other introductory Shoal article
•  GlassFish podcast interview about GlassFish Clustering and Shoal

Sunday Nov 11, 2007

JMS - JNDI, Administered Objects and High Availability

Taipei 101

Linda has two new posts; one builds on her popular JMS 101 introduction; in the Follow-up Entry she expands the hello world example to use JNDI and administered objects.

The second entry is more advanced and covers High Availability in JMS. Linda talks about brokers, clusters, and stores in the OpenMQ implementation and points to the future directions for the project.

The OpenMQ team is always interested in your comments at USERS@mq; the binaries are included in GlassFish and separately at their Download Page. Enjoy!