New on LowEndTalk? Please Register and read our Community Rules.
Why is the kernel stuck at 2.6 in OpenVZ?
So v3.7 just made its way to stable a few days ago so I'm just wondering why OpenVZ doesn't even have/support v3 yet. No complaints, just curious for the technical reasons of this. Also, yes, I know that it's rare to even see above 3.2 in modern distros (I think Debian Squeeze is still 2.6 but Wheezy is 3.2).
Because they only support RHEL kernel which is 2.6.32
Because the changes between the OpenVZ kernel and the stock kernel are quite big and it is not a trivial task to port everything to the 3.7 kernel.
I certainly wouldn't want my VPSes on a server with such a new kernel. Old == stable.
This makes sense. I also did some quick Google searches and came up with a pretty satisfiable answer. From forum.openvz.org/index.php?t=msg&th=10560&goto=48592&#msg_48592:
I suppose this answers my question perfectly: I was getting a bit antsy because I always love the latest and greatest stuff and it also means that eventually in the next few years (1 or 2?), we'll start seeing v3 kernels (3.6 at that!) in our servers!
Latest and greatest, man. Latest and greatest.
Recent OpenVZ kernels have serious I/O issues on low RAM containers. I hope they fix those before updating any further.
Latest doesn't necessarily mean greatest, especially with the kernel. All it really means is that Linus trusts that the developer has a good reason for doing whatever he's doing, and that the kernel still compiles.
On my KVM servers, I use Ubuntu Raring with kernel 3.7.0-6.
Come at me.
Would the 2.6 kernel of OVZ be capable to support newer distro? E.g. for next year, there will be newer ubuntu and debian, etc.
I have noticed that Ubuntu is quite good for virtualization as they have good repos.
Kernels that have the second major number uneven, eg. 3.7, are development kernels only. Notice how only 3.0, 3.2, 3.4 and 3.6 are maintained?
Anyway, there absolutely no point or gain in porting OpenVZ to a newer kernel.
Also, OpenVZ will never be accepted in the Linux kernel. KVM is native Linux virtualization and support for Xen was accepted as part of the Linux kernel starting with version 3.0, however since OpenVZ does not bring anything new or useful to the table, Linus Torvalds and others will never accept OpenVZ as part of the Linux Kernel. In their eyes OpenVZ is a hack, and in my eyes s well. The only true virtualization so far on Linux is KVM. Heck, as we sleek virtuos drivers are being developed for Xen. I still don't understand why OpenVZ is used by so many when KVM is mature, available and just as easy to deploy.
Then tell that to hosts. I usually see KVM plans are more expensive than OpenVZ.
In my experience using, OVZ are so fast. So I love it.
Always latest and greatest for the most furious and fearless VPS warriors
We're so bad-ass we deploy right to production!
Another thing to consider is that the RHEL kernels aren't stock 2.6.32; there's a LOT of backporting done from newer kernels into it. Not sure why RHEL chooses to go that route.
We run kernel 4 on our nodes. Hold on let me recompile before you check that...
@gubbyte shouldn't it be ”get at me” as Tom Haverford would say
Come at me bro!
@Damian - RHEL 6 was released only 4~5 months after the release of Linux Kernel 3.0. Since it's an Enterprise class product, and final versions are frozen at least 6 months in advance they don't make last minute drastic changes to the distribution. That is why they have to backport code from future kernels, however they limit the backports to drives, bug fixes and security fixes.
I wouldn't call recent RHEL 2.6 "old". The whole thing with new kernel rapid version numbering makes it looks like a lot of things have changed since 2.6.32. But if they've keep the old version numbering, recent kernels would be something like "2.6.48" only.
Besides as already mentioned, RHEL kernels include a lot of backports from 3.x branch.