The decision was made earlier this week to move forward with a new ROCKS release based on ROCKY Linux thus the new name ROCKY-ROCKS. We have forked the existing ROCKS repositories to begin work, but that is all the further we have gotten.
The project repositories are housed on github at https://github.com/techshepherdorg if you are interesting in assisting with the development shoot me an e-mail or request developer access to the repositories on github. I am happy to accept the assistance.
There have been a large number of people on the Rock Cluster distribution list asking about the future of ROCKS after the changes to CENTOS. I would like to start the discussion on a blog type forum in order to get a feel for the number of requests and expected features as well as interest of developer interested in keeping the project alive.
If you could leave a comment below with your contact information, level of interest (i.e. developer/user), and thoughts on which underlying OS to use for the future (i.e. Rocky Linux, Debian, OpenSuse, Centos). I am making no promises, but would like to see the project move forward and I am willing to take on a lead in making it happen.
You can also e-mail me at shamilton@techshepherd.org.
Thanks
I love the idea of Rocks 8. Unfortunately, I wouldn’t be much of a contributor, since I’m just an attempted user of Rocks.
I learned about Rocks, nearly two months ago, while trying to set up a cluster for running Fortran 2018 parallel programs. Once I read about the general philosophy of Rocks Cluster Software, I instantly became intrigued. I love their basic philosophy of making cluster management simple.
I’m far more of an engineer and scientist, than a computer programmer. I learned Fortran for the implicit purpose of parallelizing my algorithms, thereby making them faster and more capable. I only set out to build a cluster, in order to scale up the capability of my computer hardware. I needed hardware that can tackle the projects that I encounter. So, for me, getting a cluster to work, is merely a necessary evil to running my programs, so that they can crunch data.
I have mentioned that I’m an attempted user of Rocks. I use the word “attempted”, because, although I discovered Rocks almost two months ago, I’m still struggling to get it completely installed and configured.
I love the simplicity of how Rocks is supposed to install, in theory. Even though I ran into problems getting the frontend to install (I repeatedly encountered error messages), the installation instructions are pretty simple; which I love. I also like how simple the instructions are for adding compute nodes. It’s this simplicity that has gotten me the most excited about using the software.
For me, one of Rocks’ biggest downsides, is the CentOS operating system. I’ve found this to be a problem, even before recently learning that there’s now an issue with going beyond CentOS 7.9.
When it comes to Linux, I’ve always used Debian, and Debian-derivative operating systems. Switching to CentOS wouldn’t have been so bad, except for the fact that it seems to be reluctant to offer newer software packages in its repository. CentOS does this in the name of reliability and stability. The problem is that, in sticking with older versions of these programs, it becomes a real nightmare when you’re trying to install software that has basic minimum system requirements. Those requirements are that gfortran be newer than 4.8.5, for example.
I’ve compiled newer versions of gfortran, from source. However, I have since been running into issues with CentOS using older libraries; which I’ve been consistently advised not to update. All of this makes moving forward, extremely daunting for a non-professional programmer, like me.
If you’re going to move forward with Rocks 8, I strongly suggest using Debian as the base operating system. Not only is Debian extremely stable, it also provides updated repositories with new releases, making it easy to update. If you update the roll, you update gfortran. Also, if your target audience consists of scientists, engineers, and those who aren’t computer wizards, then Debian continues to make sense. This is because Debian is the backbone of Ubuntu, which is the most popular Linux operating system for non-computer gurus.
I love the idea of Rocks, and would be thrilled to see a Rocks 8 — especially if it uses Debian as the base operating system. Also, if anyone ever chose to provide an OpenCoarray roll, doing so would allow Rocks to provide instant parallel capability for Fotran users, which would be fantastic. For me, such implementations would be a dream come true.
Steven,
Great News, I have accepted an offer of hardware for developing ROCKS 8. We are going to be basing the project off of the existing NPACI ROCKS source code, much of which has been in active for several years. We have forked the ROCKS repositories on github and will begin shortly to advertise the beginning of the development on the ROCKS cluster list looking for additional developers to join us.
Thanks for the support in saying a new version is necessary so many have moved to things like xcat and ansible to manage their systems that I was not sure the project could be revived.
Thanks,
Scott