New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Comments
proxmox branch of openvz must be a lot more stable then, no issues here.
I think LXC and Virtuozzo are pretty much joined at the hip. I think the Virtuozzo guys are on the LXC working group. When it makes sense they incorporate LXC features. Unlike companies like Ubuntu, they seem to use whatever software makes the most sense. Not whatever lets them try steal the most marketshare or just so they can say that it's their own creation.
Depends what your users are hosting as well.
The 100 branch had multiple breakout exploits and they reintroduced the issue more than once and at least once or twice they didn't document fixing it, it simply patched right away.
Francisco
Proxmox 4 doesn't have OpenVZ anyway, it has LXC.
I had some KVM nodes set up and now slowly migrating them over to OpenVZ. KVM was just too much administrative hassle compared to OpenVZ. My actual real world first hand experience is that it's far far far heavier on resources as well. I can get twice as many servers running OpenVZ with a better user experience on top of that and no there is no overselling going on.
For one thing each OpenVZ container does not run it's own kernel so that is a big resource saver right there. Also I notice the I/O latency tends to get kind of high on KVM as you add guests as opposed to OpenVZ which acts about the same. Did all the optimizations possible. Even updated RAID1 to RAID10 which did almost nothing to improve the situation. I think part of the problem there is that individual guests can run swap whereas OpenVZ can't. I have also found some people adding 4+ gig swap files to try squeeze more out of their servers at the expense of everyone else.
I think there have been improvements in KVM on CentOS 7 but too little to late for me.
And a big administrative resource saver for KVM. All those tickets "enable this module", "enable that module" like @Francisco said
yea, that's the annoying one
Depends on what you are doing. If the stuff you are hosting needs a lot of kernel modules them maybe KVM makes more sense.
It's a non-issue for me compared to the extra administrative burden of KVM.
Probably depends how much per hour is worth the labour for the tech that will answer the support ticket vs price of the extra hardware needed to run only KVM (but if you get read of the cheaper Ovz plans then it has to be win win)...
Right, but is LXC production ready to be used as widely as OpenVZ currently is and providing as much isolation? It doesn't seem like it yet...
Kinsta recently announced they are now using LXC to power their managed hosting.
From the LXC site: "LXC is production ready with LXC 1.0 getting 5 years of security updates and bugfixes (until April 2019)."
Try it yourself. I just downloaded and installed onto VMWare. If you have used Xming/virt-manager before it works the same way to get a GUI going on your windows desktop. The virtuozzo ISO install is completely automated and sets up the br0 interface for you. I didn't have to do anything and now have a container and virtual machine running on the same server in less than an hour.
I had to
yum install xauthand then reboot to get virt-manager working and then had to manually add virtuozzo virtualization via the virt-manager GUI. That is all virt-manager and OS related and no so much virtuozzo related.https://docs.openvz.org/virtuozzo_7_beta_homework.webhelp/_using_virt_manager.html
I was also able to create containers with vzctl so looks like you can still do it that way as well. A lot of people use virt-manager/libvirt so this lowers the barrier/learning curve for them to start using OpenVZ and/or continue to use KVM. If you are used to using vzctl then it appears you can continue to do it that way. That should help make control panels like Solus compatible as well.
I hate very much OVZ because of outdated kernels.
2.6* kernels are sucks.
3.* kernels suck too, but they already start to move to better features out the box.
4.* kernels are awesome, and I like them very much, because of tons of improvements in network level, of supported devices, etc. The performances in network aspect is very different between 2.6 & 4.* kernels. 4.* Kernels much after.
What extra devices do you need to support from new kernels on your OpenVZ VPS? Name one specific feature from 4.x kernels that you need in your OpenVZ VPS. Bonus points if you don't have to go google an answer.
My intent being to separate complaining for the sake of complaining and legitimate needs left unmet.
The average person doesn't know why they upgrade kernels, I propose. Generic ideas about maybe security, maybe performance, I doubt a fraction read any kind of notes for each release.
@jarland - well ipset is a big one and as far as I can tell OVZ 7 doesn't have it supported either...
Francisco
Granted you know your kernels and you're well above the average user in my mind
Translation, the version number is higher therefore more SHINY. Must be faster with a higher verson number so you just said that right? Any proof to back it up?
They said that VZ 7 will have ipset support in the container. There is a bug report on that if you do a search.
https://bugs.openvz.org/secure/Dashboard.jspa
So it should be in there unless something has changed or they have not added it yet.
Oversell i guess, it's the only possible future i can see now for OpenVZ. LXC're really kicking VZ's ass.
How?
OpenVZ kernel is too old, and it's no longer developed, it's also have been removed from Debian.
However OpenVZ do have better feature that's more suitable for business use, example: resource accounting and isolation, that's also what i think LXC needs to improve on.
eew my eyes **** i'm eating hot cheetos makes me sick can't stop think about the *** you posted ... ****
Containers have issues in general. Both OpenVZ and LXC have had issues with SystemD based distributions. It took a while for Debian 8 to be supported on OpenVZ and they required even more kernel updates between CentOS 7 -> 7.2 due to a SystemD upgrade.
LXC didn't get out of it unscathed either with SystemD based containers taking a while to be supported.
It's working for now but for how long? It's only a matter of time before the SystemD developers hit the meth pipe again and come up with their next idea.
Francisco
Too old for what? What isn't it doing for you that LXC is? It is still in development. I don't know where you're getting your information.
Not implying OpenVZ is better, just enjoying the humor of the majority of OpenVZ hate being based on opinions that have no backing. Not that there isn't valid reasons, just that many don't seem to actually need real ones.
Sometimes, up-to-date version number can make one happier. Just like you can finish everything on a Windows laptop, why would you choose a Mac. Feeling doesn't matter for some people, while matters for others. Seriously, some people really have up-to-date OCD, and I might be one of them.
Thanks for letting me know. I guess the CentOS 7 and Debian 8 containers I have had running for months is just an unexplained phenomenon that probably has something to do with space aliens or something.
I get so much quality info around here.
This is at least the second time you have tried to spread half truth FUD just in this thread. You are like the Fox News of LET.
lol
I don't expect that OpenVZ will die soon or later. OpenVZ is still widely being used by a ton of hosting companies - something you don't change on a short term.
LXC looks promising, but I don't see the LET community move to it on a short term since there is still a lack of control panels like SolusVM which are supporting this virtualization technique.
Looking at the stats of OpenVZ, the amount of reports/new hosters looks pretty steady;
https://stats.openvz.org/t.php
General stats including populair kernel versions;
https://stats.openvz.org/
Or you could read the changelogs? initial SystemD support wasn't added until 94.7 and the latest CentOS upgraded to SystemD 219 which led to a multi month gap ( https://bugs.openvz.org/browse/OVZ-6384, assuming you're a host that sticks to stable branch) where users couldn't yum upgrade w/o breaking their containers.
Now, a host would be forced to reboot for this because kernelcare didn't get it back ported till August 2015 ( https://www.lowendtalk.com/discussion/61825/upcoming-kernelcare-openvz-patch-available-ms-hrtimer-backport-clock-boottime-feature ).
Francisco
You googled a ticket from July 2015 that was closed on July 2015 so case closed end of story I guess. You win the internet. New software not working right away is totally out of the ordinary. Bugs coming up occasionally is also totally out of the ordinary apparently. Your arguments are compelling I gotta admit.
What relevance does your time teleportation FUD have in the here and now?
Kernel 3.13+ (3.5) (2014 year) - http://people.netfilter.org/hawk/presentations/devconf2014/iptables-ddos-mitigation_JesperBrouer.pdf
Kernel 4.4 (2016) - http://kernelnewbies.org/Linux_4.4#head-7c34e3af145ac61502d1e032726946e9b380d03d
And last: http://kernelnewbies.org/LinuxVersions
Check each version above 2.6, and tell me ANY reason why 2.6 kernel is not outdated?
There are tons of improvements. I forgot to add benchmarks related to new kernels, because I forgot a link for evernote post, with tons of links & proofs.
So a newer kernel has more stuff (most of which will never be used or is not necessarily relevant for hosting). Ok got it. Didn't realize that. What about stability? Is it not a priority for you in a hosting environment?