[ovirt-users] Re: Oracle Virtualization Manager 4.5 anyone?

Hi Gianluca, nice to meet you again! Xen is in dire straits. Which is a shame, given its history and its sometimes unique qualities e.g in the area of Library operating systems... The biggest technical issue I see is x86 architecture deficits, but in the mean-time it's simply eco-system size and some choices like OCAML that have turned into one big technical debt that nobody is going to pay. I think the Xcp-ng guys are heros, on older (supported) hardware Xcp-ng runs like a charm, but there are on an extiction branch of virtualization evolution. Not a fight I want to pick, but Google fails me.. and I don't think it's Google's fault Without a roadmap and committed life cycles, nothing is a product. And with oVirt lacking both, Oracle's "fork relation" is nothing to build on. It breathes life into what is potentially a carcass with lots of potential... But for how long? It was Redhat who spread the fiction of GlusterFS as the one solution without any scaling limits. Your opinion seems to resonate even with Redhat, who decided to discontinue any commercial product based on GlusterFS. But I need HCI, and Redhat/Oracle/oVirt isn't providing an [integrated] alternative. Ceph seems ok for that, but nobody is working on instrumentation. I've round Oracle Linux sources on Githubu, but their fork of oVirt is nowhere to be found... I've seen that, and it looks like it's being monitored. But the equivalent for oVirt or OV is missing! I'd be more than happy with QA lasting a year: I'm not going to deploy leading edge hardware for hosting VMs. What I still need is a roadmap and comittment. I'll try, but I'd really need the oVirt equivalent. One of the negative experiences here has been that any issue that was KVM, Gluster, VDO, Ansible, Cockpit or just RHEL wasn't dealt with here: they were entirely focused on oVirt and just what *was* oVirt wasn't immediately recogniseable to a newcomer. And in a *product* it should not matter anyway... Not so much an understatement we don't see any more from the likes of a Boris Johnson or a Nigel Farage... I'd say "total information blackout" seems a better fit! Considering that Oracle's oVirt is an oVirt fork and they are at least referencing oVirt on their web-site: there has been zero participation of Oracle on this forum. It's a critique purely aimed at Oracle, certainly not at the Redhat guys doing most of the oVirt work! That's just it: to my understanding the 100% interface compatibility has been a major UEK selling point, personally underlined (and signed in blood) by Wim Coekaerts. That's why I was quite annoyed to find that switching kernels wasn't transparent and that one of the first things I discovered upon trying to migrate from oVirt 4.3 to OV 4.4 failed for lack of VDO support. Even Oracle's RHEL kernels sometimes failed to keep VDO support working, which created huge issues as I was blindly applying patches around Christmas 2022 or 2021: I don't tend to forget having to undo updates when I should be singing Christmas carols with my kids! Again, what I need is an OV community on top of UEK. Should even apply to non-paying ones, like me! To me it's one of the pretty near 'criminal' offenses of oVirt: they advertised oVirt as a solution "for your entire enterprise" and that included HCI (using GlusterFS as the *only* storage based for that). It took me years to understand that the Israeli KVM, SPICE and oVirt teams originally launched by Moshe Bar would not look or comment on any issue cause by KVM, VDO, Ansible, GlusterFS or just Linux, because it was outside their "purvue": how would anyone who hat just followed the promising jargon on the oVirt home-page know what's from which Redhat acquisition? GlusterFS was a miracle in terms of limitless scaling nad oVirt HCI integration... until you had an issue or it was siltently sidelined. That may not have been an oVirt team decision, but imagine Linus Torvalds suddenly disowning file systems or claiming that any virus or trojan propagating via Linux APIs wasn't his responsability, unless you paid him personally to make it that. Open source is built on trust and if you don't take your responsibility seriously, the damage to the eco-system is "significant" to say the least. Redhat under IBM leadership has made some significantly bad turns. Perhaps I do, but the marketing around UEK is that it's the same, only much better. Just as Redhat promises API backward compatibility for any major release kernel, UEK promises that UEK is like the Redhat kernel, "only better". That's Wim's message, not mine. Admittedly it's been a few years since I heard it from his own mouth and we shook hands on it (and he also left Oracle for some time in-between), but I consider that as much of a guarantee poured into concrete as you'd ever get in IT. More than happy to have done so and I'll continue my testing as well: I really do wish oVirt well, even with all the huge disappointments I've lived through with it. I have kids, disappointments and loyalty somehow do not seem to correlate. It is much, much simpler, very straight forward and it just works. Albeit with its functional limitations. It's extremely quickly to set up and operate and the main magic is in KVM and QEMU, so it can't be bad and a week-end may be sufficient to explore it rather well. Ciao e vi auguro un meraviglioso Natale! Thomas