TL;DR - It’s Friday (the 13th ! ZOMG), I did lots of work, and you, as usual, deserve a beer
Central Operator on Duty (COD)
Things are getting better on the grid infrastructure, with ZA-CHPC bringing services back up and ZA-UJ fixing several at their site. There was a Trust Anchor update to Release 1.62-1, which required all sites to update their installation. Only minor issues were had. A major problem however was the downtime of the Open Nebula cluster at ZA-MERAKA, which hosts the core and ROC services. I opened a downtime on the GOC to announce this, and hopefully we can salvage some A/R, but this just highlights again the need to finish our Core Services Failover Campaign
Several new jobs have been registered in the SAGrid Continuous Integration platform, gearing up for the GADGET deployment. We’ve also had a request from a user at UJ for Quantum Espresso. Applications are deployed into the CVMFS repo, which is in beta, but looking good.
We’re about to start a H2020 project called Sci-GaIA which will be deploying data repositories across Africa, amongst other things. Gearing up for this project, I’ve worked on an Ansible role for the Zenodo/Invenio stack. This provides an OAI-PMH compliant data repository out of the box, along with DOI integration (at least, in theory !). This turned out to be a perfect place to start using something I’ve been interested for a while : Docker.
I’ve been working with Docker for the last few days and it’s turning out to be a very useful tool for development. Together with Ansible, it’s a great way to develop playbooks for the configuration of services (as long as they’re on a single host - I haven’t figured out clusters of containers yet.)
DevOps Release v0.0.4
I’ve opened the pull request for the next pre-release of the DevOps - v0.0.4. This will have the IdP (backend and web interface) roles included. The release will be accompanied by a DOI from Zenodo and a bit of a writeup. The CHPC is using the code to redeploy their worker nodes and user interface, which is tied to Perun on the fly (which is, how you say ? “Very Rad”). We’re one step closer to fully integrated identity management and service deployment.
Our Ansible brings all the sites to the grid !
The CHAIN-REDS final conference is coming up at the end of the month. I will be representing the Sub-Saharan region, and making a demo of the Executable Infrastructure we think is so cool. The demo will show the deployment of an integrated LDAP/IdP and Science Gateway with Ansible.
I also invited the H3Africa Bio Net project to the conference, which will be funded by the project. This will be a great opportunity to cross-pollinate ideas and needs.
The project is running as one of it’s final deliverables an HPC survey, coordinated by GRNET. We need to get the feedback of all of the “HPC” centres in the region, so that we can construct the state of the art of the region regarding usage and adoption. This has been started. Of course we expect input from the CHPC, but more importantly also the newer sites in Tanzania, Zimbabwe, Kenya etc.
During the CHAIN-REDS project, we worked with the African Population and Health Research Centre, in order to improve the visbility and citability of their data. This involved using handle service hosted by the European Persistent Identifier Consortium (EPIC). We decided to deploy a handle service in South Africa to act as a catch-all for the region, in order to allow data repositories like APHRC’s to become more visbile. GRNET is going to help us install the handle server, however, we need to buy a prefix from CNRI which hosts the registry. These cost 50 USD a year.
I’ve exchanged a few mails with folks at UCT and CHPC about re-writing the SAGrid Strategy in a post-NICIS world. This has had very positive feedback, and is currently evolving quickly. I wrote my extensive thoughts on what I would like to see, so if you have any comments, head over there, read the fine article and make your point of view known.