How is it possible the VPS owner break into main node?
I am shocked. Is openVZ really safe at all?
Anyway I can check how it was done and how it can be avoided in future?
Can anyone help me with this??????
Thanks.
[Updated on: Thu, 19 October 2006 04:28] by Moderator
John Kelly Messages: 97 Registered: May 2006 Location: Palmetto State
Member
From: *isp2dial.com
whatever wrote on Thu, 12 October 2006 13:54
How is it possible the VPS owner break into main node?
The VE has a chrooted, limited view of the filesystem. So the VE cannot "break in" to the HN filesystem.
However, the VE can use ssh to login to the HN, just like any other networked host. Maybe you have poor security on your HN. But how can we know what happened, when you don't provide detailed, factual information?
How I detected?
In hardware node we use alert script whenever anyone login to root we get alert. And direct root login to Hardware node is disabled. To get root access one has to login as user allowed list in hardware node and then su password to get root.
There are only 2 users in hardware.
The VPS user got the access to root. And the details of VPSuser, ip, time etc were recorded in alert email.
This happened 2 times with different VPS users.
I can send the VPS root access and alert details to the developer of openvz to have a look at it.
Maybe they can understand better then me.
John Kelly Messages: 97 Registered: May 2006 Location: Palmetto State
Member
From: *isp2dial.com
whatever wrote on Fri, 13 October 2006 12:01
And direct root login to Hardware node is disabled.
Maybe. Maybe not. It's possible there could be an OpenVZ bug, but it's more likely your login security is misconfigured.
If you can identify a real bug, the OpenVZ developers will fix it. That's their job. But they don't have time, and it's not their job, to train OpenVZ users how to manage system security.
jason|xoxide Messages: 20 Registered: September 2006 Location: Exton, PA
Junior Member
From: *phlapa.fios.verizon.net
Well, if it's a kernel bug that only affects 2.6.17 and below then it will go away as soon as the test kernel is migrated to 2.6.18 (which they already said would be soon). You can't very well blame OpenVZ for something that is also broken in the vanilla kernel.