If they just leave CentOS to languish without proper alternative that leaves a pretty big hole in the market that someone enterprising could fill, potentially really hurting Red Hat in the long run.
Oracle seems like a reasonable candidate here, if they throw their weight behind a community fork of CentOS (like Rocky Linux), or create one themselves, that could shift public perception rather in their favour, I imagine.
Still, big investment, somewhat risky, and Oracle and risk…
That’s literally what Oracle Linux is. I mean I guess it’s technically based on RHEL, but it’s close enough that you can easily convert CentOS to Oracle linux.
I’m aware of Oracle Linux, and what it is (among others, one of three distros that still run on UltraSPARC).
It’s not community maintained though, is it? If it is they’ve managed to hide it well at least.
In the past Oracle could just point to CentOS if people wanted to “see the code”* (barring the kernel modifications, I guess), but that won’t really be the case anymore, hence the “void” I mention.
*by which I refer to the the way the packages are built, which was CentOS’ entire claim to fame, not the actual application source code
I’m aware it’s a RHEL fork. CentOS got big by being a community project that attempted to stay compatible with RHEL (at least until after RH annexed the project).
Being open leads to quite a bit more trust, certainly when companies like Oracle are involved, not to mention community goodwill. Corporations probably don’t care about the latter too much, but being pushed by employees is how most big FLOSS projects get a foot in the door in the business world, including Linux itself, and being blatantly anti-community tends to be how they die, or at least fade into insignificance (Oracle knows a thing or two about that ). Hell CentOS just being there probably helped RHEL get accepted in places it other wouldn’t have been.
If Oracle wants to capture mindshare then working with the community would be to their benefit, certainly given Oracle’s reputation, and likely to Red Hat’s detriment, since they’ve been doing “everything but” especially when it comes to the entire CentOS thing.
Yeah definitely agree with that. Oracle could just as easily pull the rug out at some point.
I’m going to wait to see which way the appliances go. oVirt in particular. I’m not sure if they can pull off a complex hypervisor on stream. I tried to install it manually the other day and I couldn’t get it to work. I think they rely on a lot of consistency in package versions.
Not sure what features oVirt has, since I never used it and was never curious enough to look beyond what it is in a few minutes. But if all else fails, you could try OpenNebula (it’s also built on top of KVM with libvirt). We used it in my company in the past. It was quite quirky (if you weren’t careful, you could delete a VM you did not want to, but we found a workaround by clicking the VM itself and not do a quick action from the selection menu) and it had too many features that didn’t help us in particular, like cost control, but for a DIY private cloud, it basically competes with OpenStack (to some degree) and it’s easier to implement. But depending on the size of your cluster, it could take a little to migrate everything; it was easy for us to migrate from OpenNebula to Proxmox, we could have automated it, but it only took 1 person, ie me, just under a week to recreate the VM configs and move the disk image files to the new locations (which were on the same NASes).
I still have nightmares with OpenNebula, but now that I think about it, I probably didn’t give it a fair chance back then (and neither did my colleagues), not to mention it had lots of updates since. Our implementation was also kinda bad, since the Orchestrator was running on a virtualization host, instead of a separate server (and if the host would ever go down, it would take the orchestrator with it, however, the VMs on the other hosts would keep running and even be accessed via Virt-Man).
I’d probably go with Proxmox, although maybe xcp-ng depending on how they divorce themselves from centos (although they’re still on 7, so probably won’t make that decision for a few years).
I wonder how this fares for the cPanel project. They were quite upset about the switch to stream. Crossing fingers they will support RHEL going forward.
Idk if there are any legal issues with using oracle in an appliance but that still seems to be the best solution for them.
Curious what this means for FreeIPA, oVirt, OpenShift, etc since RHEL already has them in rebranded form. I guess they will just be unstable versions in stream.
Also, why didn’t they announce this at the same time as the CentOS news? Would have saved them a lot of bad press.
Yep, just like RedHat intended CentOS Stream to be. Well, again, I believe “unstable” is a bit of a misnomer, it should be pretty solid, but again, not “Red Hat solid.”
Apparently Mattermost and AWS became Rocky Linux sponsors:
There are so many moving parts in ovirt, openshift and FreeIPA, they really rely on locked package versions or things will break. CentOS stream should be stable as an OS but these big services might not be.
Does it work on stream? I read somewhere that you can’t lock packages in stream, but idk if they were referring to that or just the difference between the rolling vs point releases.
once you have the iso… you can do as many as you want…
so what they’ll do is monitor how many systems check in with subscription manager from the same ip(or acct login)…
and if you hit the 16… then someone might be reaching out or sending a letter/invoice…
). Hell CentOS just being there probably helped RHEL get accepted in places it other wouldn’t have been.