I'm having flashbacks to the Fed4Fire+ experiments we did in 2021.

Brett Sheffield (he/him)

I'm having flashbacks to the Fed4Fire+ experiments we did in 2021. We worked right through Xmas - the testbeds were quiet that week with everyone on holiday so we hammered the network while we could.

It was difficult to get enough nodes to run the larger tests, and several of the nodes were actually broken.

We had to first start an experiment to request the broken nodes to exclude them from the pool, then start our experiment before releasing them back to the pool.

Brett Sheffield (he/him)

There were a bunch of accessibility problems too. One of our team has low vision and couldn't use the java frontend and the cli API was largely undocumented.

We reported the problems but were told rather rudely that if the network didn't suit us we could just give back our funding and go away. We'd already invested far too much time and effort at that point so we ploughed on but it was an incredibly stressful experience.

Brett Sheffield (he/him)

There was no induction process. In fact we were never told what testbeds we had access to. We found out several months after we'd started that we had access to other testbeds but no one had bothered to tell us!

It was only when we were reporting yet another bug in their systems that someone from the other network replied and asked if they could help and it dawned on us we had this other network to test with. Too late to make any use of it, unfortunately.

Brett Sheffield (he/him)

The experiments were still useful, but we could have done so much more if we'd known ahead of time what we were working with.

There's so much EU money that's been poured into these testbeds, but the management of it is utterly shambolic.

I'm told this is "normal" in academia ‍️

Brett Sheffield (he/him)

Even the website for Fed4Fire+ is now defunct which had the interviews with teams etc. It's not that the domain has expired. No. The marketing company that ran it decided to redirect the domain to their own website.

‍️

Brett Sheffield (he/him)

Correction: the website is back up. It was down for more than a year, as verified by archive.org. Did someone give them a kicking when I wasn't looking?

Esther Payne :bisexual_flag:

@dentangle perhaps there's a good reason to ressurect the initiative. For one thing the threat to the overall NGI funding. Either that or Fed4fire is planning a launch of a new consortium under NGI.

Although for 2025 there's no new budget at the moment for new funds to launch. So showing all the different initiatives under #NGI is wise.

Librecast - Decentralisation and Privacy with Multicast

(www.librecast.net)

Brett Sheffield (he/him)

@onepict My understanding is it has been replaced by SLICES-RI, which appears to be the Fed4Fire consortium in a new trenchcoat. Mostly the same labs under a new name.

There are some useful resources there but I'd urge anyone applying to use the labs check *very* carefully what is on offer before committing.

Some labs are ancient, broken, appallingly managed, have poor accessibility, and you could run more useful simulations on a modern laptop.

And some, I assume, are good people

Esther Payne :bisexual_flag:

@dentangle I hope the honest feedback I gave in the feedback questionnaire and the presentation review of our results helped the consortium improve.

But there is a need for a bank of high availability testbeds that can test for real scale. The consortiums also need to be more setup so that individual and non incorporated projects can apply.

So I hope @ngi are considering how that could be funded. The @REA does need some frameworks for this type of research at scale

Testbed Federation funding for FOSS Projects

We discovered that grants like fed4fire+ have been set up to only allow Companies that are VAT/TVA registered to be eligible for funding. This means that non-incorporated FOSS projects and individuals aren’t eligible fo…

Discuss Social Coding (discuss.coding.social)

Esther Payne :bisexual_flag:

@dentangle Perhaps looking at other frameworks for federated testbeds to access computer labs would be wise as well.

So perhaps more under a transatlantic or even global scientific research cooperation.

Like CSIRO in Australia, or labs in Africa and South America.

Brett Sheffield (he/him)

@onepict

> "it connects with grid 5000 using global vlans. Which if you aren’t careful will knock out the backbone."

Ah yes! We created some nodes for testing, but the documentation was rather sparse and so we did not realise our test nodes were not, in fact, on the same switch but were actually on opposite sides of the country. When we launched the test we saturated their backbone.

We could have done some very interesting things with Grid5000 if we had known we had access sooner.