Friday Feb 15, 2008

Recent Open HA Cluster Activity

There have been some exciting developments in the Open HA Cluster open-source community over the past couple months. Specifically:

Nicholas Solter, Sun Cluster Developer and OpenSolaris HA Clusters Core Contributor

Tuesday Dec 04, 2007

Sun Cluster Geographic Edition is now Open-Source

The source code for the Sun Cluster Geographic Edition product is now available in the HA Clusters community on! In addition to browsing the Open High Availability Cluster Geographic Edition source code, you can download it and build it with either the Sun Studio or the gcc compiler.

This source code release represents the second phase of the complete Sun Cluster open-sourcing roadmap. The first phase, the Sun Cluster Agents, occurred last June, and the third and final phase, the Sun Cluster core gate, will happen sometime next year.

I'm particularly pleased that, in addition to product code, this release of the Geographic Edition source includes test code, man pages, and globalization source.

Nick Solter

HA Clusters community facilitator and
Sun Cluster developer

Thursday Nov 29, 2007

Sun Tech Days in Rome and Milan, Italy

Sun Tech Days are a series of developer focused events being organized world-wide by SUN. The main focus of these events are Java and OpenSolaris. The emphasis of the events is less on product features and more on engaging developers to work with the product. Here is the main event link

There were a couple of "tracks" (series of talks with a specific theme) in Rome and Milan: OpenSolaris, NETBEANS, Java and Solaris Development. The Java/NETBEANS related tracks were going on in parallel with the the Solaris ones. In addition, in Milan, there also were some booths where people can come in and looked at demos, get Solaris installed on their laptops etc. This page has the Agenda in Rome and Milan

With Solaris Cluster, our talk titled "Using and Contributing To Open High Availability Cluster" was part of the OpenSolaris Track. It also included a demo of Cluster Express running on a laptop running OpenSolaris and demonstrated failover of an application (Apache) between two zones on the same machine. Here are some of my impressions of the events:

OpenSolaris Day

OpenSolaris welcome was done by Chris Armes, Solaris Marketing director. Chris made a few jokes to get the audience to start paying attention (he seem to be good at that!) and also pointed out a bunch of goodies (T-Shirts, Books and Solaris Cluster demo DVDs), available for the audience members.

There were several presentations before Cluster, covering OpenSolaris, Nevada plans, Solaris Virtualization strategy and Solaris Networking enhancements. Virtualization topics generated most of the comments and questions from the audience, with many people asking about the plans around xVM and OpenSolaris.

My presentation on ""Using and Contributing To Open High Availability Cluster" was scheduled after the networking presentation. Before starting the presentation, I did a show-of-hands and not many people in the audience had prior knowledge of Open HA Cluster.

Given that, I spent a bit more time then I had planned, on introducing the basic concept of Availability and Solaris Cluster itself. I talked about concept of SPOFs and how Sun Cluster deals with failures in Storage/Network/Server/Application etc. I moved quickly thru details of what a SC Agent is and what it does. Next I showed a demo of SC with Apache as the application failing over between two zones on the same single node cluster (my laptop). The demo used the "keno" demo which continuously contacts the server and depending upon which node it is, it displays a square of different color. After failover, the client application starts showing squares of different colors. I demoed killing the apache daemons and quick restarts of the application as well as halting of the zone etc. The demo was well received by the audience, although the only comment I received was to use the "uadmin 1 0" command to reboot a zone instead of "reboot" command which I had planned to use. Hey... At least it shows the audience wasn't sleeping! :-) :-)

After the demo I talked about Solaris Cluster plans for open sourcing the code. The fact that all the Agents are already open sourced and there is a roadmap for more. I closed with (and spent some time on), the slide with all the Open HA Cluster related resources. I am hoping that at the very least, the audience members would remember the OHAC community web page on

I talked a bit about the contents of the Solaris Cluster demo DVD and encouraged people to install it on their laptops. I made the point that the presentation laptop itself is running Cluster Express on OpenSolaris. Which seemed to get the point across that the Open HA Cluster is easy to setup and run on common, off the shelf, hardware.

OpenSolaris Installfest

In Milan we had a OpenSolaris booth where people could come over and get their laptops installed with OpenSolaris. I had duties on all of Friday. Thanks to feedback from Boston about lack of software to partition people's hard drives to create empty partitions for Solaris, I was ready with my own copies of the partitioning software.

Lots of people came over to the booth simply to have a chat about Solaris, not really to install their laptops with it. The most common question was "Why should I use Solaris?", we tried various ways to tackle that question such as unique features and strengths of Solaris, because it is the most reliable and scalable (I had some amount of traction on the reliable part, although strangely enough, from the POV of the quality of code). Another point which (I thought) connected with some people was: Because it is fun and new and because you have never tried it before.

There were lots of questions on whether OpenSolaris can be run via VMware, answer is yes. I took pains to remind people that if they run Solaris under VMware, don't attribute any slowness or lack of stability to Solaris, rather to VMware.

Most people were wary of actually going thru the repartitioning of their hard drives. Some which did go thru were pleasantly surprised to find that once the right partition is specified to the installer, it goes thru the install very quickly and painlessly. I think this is one area where OpenSolaris has improved a lot recently and it shows. I engaged others (not ready for repartitioning of their hard drive yet) by putting in the OpenSolaris DVD into their laptops and running the Solaris Device Detection tool on their laptops. For all of the people who tried that, the device detection tool was able to determine that Solaris would run fine on the laptop. Only missing driver was sound driver, which is not essential, but I suggested to people that opensound drivers would solve that problem as well, as they have for me. Hopefully a couple of them would try out the OpenSolaris install by themselves.

If you missed the event in Italy, check out the schedule on , you may just find one near your home town. Find out more about Open HA Clusters at OHAC community web page on . If wanna join the conversation about it, or just wanna listen to what people in the community are talking about, visit the Open HA Cluster forum, which also has a link to discussion archives.

Ashutosh Tripathi
Solaris Cluster Engineering

Friday Aug 17, 2007

Eyes Wide Open

Well this week really has been a case of eyes wide open, either the roller coast ride of the Dow or Nasdaq or the amazing IPO of VMware will do it. However, I'd argue that in these cases "eyes wide open" simply reflects the fixed gazed observer in us.

Instead, I'd argue that "eyes wide open" correctly reflects the recent Sun and IBM agreement,"IBM Expands Support for the Solaris OS on x86 Systems". In this scenario, I'd argue we are not fixed gazed observers but serious players making their moves.

Today we are releasing onto OpenSolaris, Open HA Clusters Community (OHAC), the source code for the IBM WebSphere MQ and IBM WebSphere Message Broker agents. Within Solaris Cluster and OHAC we take these IBM products seriously, ensuring high availability support for WebSphere MQ V6.0 and WebSphere Message Broker V6.0 on Solaris 10 SPARC and Solaris 10 x86-64, also within Solaris Containers.

I think it's important to try and demonstrate the purpose of OHAC. As a relatively new community, within OHAC we have already endorsed developing an agent for IBM Informix v11, please see the HA-Informix project. Once developed it is our intention to putback HA-Informix into Solaris Cluster, thereby ensuring high availability support for IBM Infomix v11 on Solaris 10 SPARC and Solaris 10 x86-64, also within Solaris Containers.

While I'm here, I also think it's worthwhile to briefly outline the steps to propose a project to the OHAC community.

1. Anyone can propose a project to the forum. Here you simply need to outline your proposal. Please see a recent proposal for an example that proposes HA-Xen.

2. The proposal then gets reviewed by anyone that's interested, but perhaps more importantly by the HA-Cluster's core contributors. Here those people will vote, either positively (+1), negatively (-1) or by abstaining (+/- 0). Please see Article VIII for the gory details although please note a consensus must be reached for the proposal to be approved.

3. If approved, you then need to submit your proposal to the OpenSolaris Governing Board (OGB). Please see an example used by the HA-Informix proposal.

4. The OGB then set you up with a project page from which you can store your design docs and general project details etc. As the sponsor of the project you will then be the project leader. Again, please look here for an example.

The next steps are then to interact with the community on the design or any open questions, develop the agent code and submit a code review, although we're in the process of providing you with a checklist of things to do.

Once all this is done, one of the HA-Cluster's core contributors will putback your code into the Open HA Cluster (OHAC) gate. Once it's there we then have the "small" task of getting it putback into Solaris Cluster. This process is not yet open and as such an internal SC committee reviews new features to be included within SC. Nevertheless, by this time most of the hard work has been done and if we believe that "the more supported agents we have then the more appealing SC is to customers" then putback into SC is just be a formality.

Finally, within Solaris Cluster we're focused on virtualization. This means working to provide solutions that use Solaris Containers (b.t.w did you know that lx type branded zones are now supported in "failover" Solaris Containers, please check out SC 3.1 patch 120590-05 on sunsolve and soon available in SC 3.2 although this is a whole new blog topic.) Other virtualization areas we're focused on in Solaris Cluster address LDOMS, VMware and Xen, although these are also further blog topics.

In fact we've recently refreshed all of our code which includes amongst many, the putback to support lx type branded zones in the failover HA-Zones agent. We believe addressing our development through OpenSolaris allows us to ensure that whatever is developed is well thought out, meets the need and ultimately gets supported in Solaris.

I'd argue most people would agree with Larry Ellison's recent quote, "Open source is not something to be feared. Open source is something to be explained. Open source wins not because it's open and not because it's free. Open source wins only when it's better."

Sounds good to me.

Neil Garthwaite
Solaris Cluster Engineering

Wednesday Jun 27, 2007

Journey to Open High Availability Cluster

I am sure you have read Jonathan Schwartz's and Rich Green's blogs about Sun's Open source initiative and plans. We in Sun Cluster were just wrapping up the Solaris Cluster 3.2 release end 2006, when Nick told me that he wanted to do some investigation around open sourcing and see what it takes to do so for Solaris Cluster. One thing lead to another and before I knew it, we started the open sourcing effort for Solaris Cluster in January this year. The project was referred to as Pleiades (the open cluster of stars closet to earth).

The commitment to deliver end June seemed like a daunting task, at first, given all that we did not know.

Six months later and here we have our first release of Open High Availability Cluster. We are committed to sharing more code in the 18 months and working with the community. I encourage you to check out our HA Clusters community. You can read the feature story, which has more details on the announcement. Also listen to a podcast with Keith White.

This journey has been short and intense! The small distributed team, located in Austin, Bangalore, Colorado, France, Germany and Menlo Park, did an awesome job coming together and helping orchestrate this first release. Yet another example that demonstrates what this group is capable of doing!

I joke with Nick that only one month after returning from paternity leave he has had a second baby this year! :-)

Sr Manager,
Availability Engineering


Oracle Solaris Cluster Engineering Blog


« July 2016