You can run Chromium from OpenVZ now?

[Geek]

What with all of OVZ's semi-recent changes (vzctl 4.9) along with the last two kernels allowing containers' access to /proc/cgroups (and other areas of the HWN) I decided fart around in QA. I set LOCAL_UID and LOCAL_GID values in my vz.conf and enabled the sysctl/postcreate scripts, spooled up two CTs with 12.04 and 14.04, installed MATE and Kubuntu, and look, Chromium!

Back when I last read about this, if you could get it to start from a container, outside of sandbox mode (or at all, really), it was essentially because of a security flaw/some sort of arbitrary access to the HWN...hence the adjacent homework in the other tabs.

Is this common knowledge now and I'm just way behind on this one? 
Do you think it's trustworthy (as it relates to the VPS itself)?
Anyone else wanna try it out on their own?  042stab106.x - 107.x with vzctl 4.9. 

Have fun,

-Edel

 

[Geek]

What the hell... worked in Fedora 21 which is still in beta.  Isn't it ironic ...the kernel that came out on 4/20.  They have a sense of humor too. 



Yeah, I started it as root.  I know... I just wanted to see if it would work. Fedora... heh, that's the first time Fedora's ever done anything for me.  Ever.

 

[MannDude]

Fedora has been making waves lately it seems, I've been hearing a lot (more) about it in recent months than usual... even though it's probably been ten years since I've last used it on a desktop.

What size VM are you running this on? I do not recall ever having issues running Chromium in the past on remote desktops but I don't entirely recall if they were OpenVZ or KVM based servers.

 

[Geek]

I've been digging into KVM the last 9 months for business reasons, and Chromium has fired right up every time I've tested a new configuration.  There are dozens of threads about it being impossible to start in OpenVZ without namespace/setuid binaries, and last night was the first time I ever got it to actually function with OpenVZ.  Usually it crapped out. Originally I thought adding LOCAL_UID=100000 / LOCAL_GID=100000 in the VZ config was what might have done it, but then I built one in production (Fedora) and it went off without a hitch.

The Fedora container is a 1GBx1GB RAM+VSwap allocation. I can tell it's hungry for more, but it functions.

 

[Geek]

Also works on my QA Wheezy node and an Ubuntu 12.04 container.  Now to find out why...