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
.
..
...
sigh
People are generally more helpful if you make an effort to spell what you need right, at least.
What do you want to know about it?
https://en.wikipedia.org/wiki/The_Legion_of_Extraordinary_Dancers
Never seen it, which is surprising because my wife usually likes things like this and forces me to watch them with her.
LXD exists It's more or less a wrapper:
https://linuxcontainers.org/lxd/
Francisco
From what I recall about reading about it for 5 minutes, it's supposedly somewhat more secure than lxc.
I'm waiting for lxe.
How about lxf?
Don't run alpha/testing builds in production.
I'm from the future, lxz is popular here.
I heard that everything between 1998 and 2016 have been officially stricken from the record books- like Germany did with the 30s and 40s.. any truth to that?
If you like buzzwords, read the introduction from Canonical: https://www.ubuntu.com/containers/lxd
It builds on top of LXC and provides some pretty advanced features like a REST api, snapshots, live migration, a image server and more. It's basically the Docker equivalent for full system containers. Unlike OpenVZ, it works with the mainline Linux kernel and has fewer limitations (nested LXD, Docker, KVM work).
I'm currently developing a project on top of LXD, and I'm pretty happy with it. Documentation could be a bit better and there's no large ecosystem around it like there is for Docker, but I'm positive these will improve in the future - eventually, people are going to move on from OpenVZ, right?
As long as you can throttle the users and get by with overcommitting resources, it will never go away.
LXC/LXD can do the same by utilizing cgroups on a mainline kernel. Not that you should, but it's possible.
Where there's a will, I suppose. The fact that although there have been "updates", that OVZ is still pretty much locked to 2.6.x forever drives people nuts, even though there are rarely specific features their $10/yr VPS will ever utilize from a newer kernel except, maybe, a slightly less screwy TCP stahahaha i make joek.
LXC/LXD doesn't have a decent way to handle storage nor sub quotas so that's a big problem. You can use something like an LVM but that still doesn't pass quota support.
Francisco
@Francisco any word on those Onion slices?
Based on this post, disk quotas are supported for ZFS and btrfs. IO throttling seems to be limited to a single value per block device, although it's not clear to me whether it applies to all containers or just a single one.
ZFS quotas aren't supported by repquota and such cPanel can't use them. It can limit the users available diskspace, but that's about it.
BTRFS should never be spoken about for anything short of pipedreaming.
Francisco
I use LXD daily as do many others.
Some get confused with LXC v1 and LXD (re LXC v2). LXC v2 has many enhancements/differences over LXC v1.
LXD can do almost anything I've tried with it from clustering, to running entire OpenStack in LXD etc.
There are alot of great example use-cases described by posts on the LXD sub-reddit:
https://www.reddit.com/r/LXD/
Also regarding @Francisco's comment about storage handling... Stephane Graber just finished posting new information to his 12 part LXD blog series on LXD about new Storage features. Francisco - check the LXD Github documentation URL below and search for "Storage". For example - "volume.zfs.use_refquota" and "zfs.use_refquota".
Storage pool configuration keys can be set using the CLI lxc cmd tool with:
Lastly...
The LXD Github has a lot of documentation related to configuration.
https://github.com/lxc/lxd/blob/master/doc/configuration.md
At the end of each "section" there are CLI examples of how to set/unset any configuration parameter for an LXD container.
An example command to set a parameter for a "device":
lxc profile device add [key=value]... where "key" and "value" are described in the above URL "device" section.
Also, if you use LXD I'd highly recommend subscribe to the lxc-users mail alias as the developers monitor that and answer Q&A daily for both LXC v1 and LXD (LXC v2) questions:
https://lists.linuxcontainers.org/
Brian
Not seeing it Got a link?
If it's the usual "BTRFS/ZFS are the way to go" then that still doesn't help people needing sub quotas (quotas for their own users like cPanel/Plesk/etc). Not unless there was some miracle change in either platform to support repquota/etc.
That isn't LXD's job to implement, that's on EXT4/XFS to figure out, just pointing it out as a blocker for many.
Francisco
It takes years for these things to mature. Everything I have read says LXC v1 is full of security holes making it unsuitable for public hosting type applications and LXC v2/LXD is still experimental.
Also, it's mostly only Canonical supporting LXC/LXD from what I can tell. Companies like Redhat don't seem to be playing along.
If I am incorrect in any of these assumptions please let me know. I don't have any bias other than wanting to pick which direction I go forward from OVZ6. So far OVZ7 still looks like the clear winner for public hosting. Been doing a bit of testing with it on Virtualizor and it seems to just work. LXC v1 also works on Virtualizor but not secure for using in a public hosting application .
I would test LXCv2 and LXD but there is no viable hosting panel supporting it yet.
Francisco the only major problem that BTRFS has is with Raid 5/6 which they acknowledge and document pretty thoroughly.
I've been using BTRFS Raid 10 on multiple 4-4TB disk servers for 3-4 years without any problems. I also use BTRFS on my single drive laptop for the past 2+ years without a single problem.
A few ZFS advantages over BTRFS:
Some advantages of BTRFS over ZFS:
You're one of the few lucky ones that are riding it out then since we used BTRFS for our backups (at the time it was mostly just rsync based) and it was nothing but a gongshow. It's possible we were an edge case due to the amount of metadata work we were doing but we must've had a half dozen full arrays lost due to one bug or another.
ZOL is mostly good but will be on point once they address their ARC issues.
Francisco