Closed yunohost-bot closed 6 years ago
Author: sbadia
Hello,
Je peux prendre ce point! Je viens de voir que la vm est en para-virtu, elle est donc sur un noyau 3.16…
Coté hyperviseur ça donne :
kernel = '/boot/vmlinuz-3.16.0-4-amd64'
extra = 'elevator=noop'
ramdisk = '/boot/initrd.img-3.16.0-4-amd64'
Je propose donc d'installer grub et un noyau dans la vm, et de passer en full-virt (utiliser le noyau de la vm et non-plus de l'hyperviseur).
Coté modification sur l'hyperviseur ça donne :
kernel = "/usr/lib/grub-xen/grub-x86_64-xen.bin"
extra = "(hd0)/boot/grub/grub.conf"
Coté vm, j'ai encore ma clé SSH sur cette vm, je peux donc faire la partie vm également.
Cela conviendrait ?
Seb, pour Gitoyen
Author: alexAubin
Salut,
ben de mon côté ca me va, t'as l'air de savoir ce que tu fais :P
Author: sbadia
Status Changed: In Progress
Author: sbadia
1. apt install linux-image-amd64 grub2
1. apt-get clean
1. apt-get autoremove --purge
1. lxc-ls -f
NAME STATE IPV4 IPV6 AUTOSTART
--------------------------------------------------
yunohost_demo1 STOPPED - - NO
yunohost_demo2 RUNNING 10.1.5.4 - NO
1. halt
1. => Modification du fichier pour utiliser le noyau/grub de la vm
1. xl create -c yunohost.srv.gitoyen.net.cfg
1. lxc-stop -n yunohost_demo1
1. lxc-start -n yunohost_demo2 -d
Author: sbadia
I'll (Gitoyen) need to update the hypervisor :)
root@yunohost:~# bash spectre-meltdown-checker.sh --variant 3 --batch json
[{"NAME":"MELTDOWN","CVE":"CVE-2017-5754","VULNERABLE":true,"INFOS":"Xen PV DomUs are vulnerable and need to be run in HVM, PVHVM, PVH mode, or the Xen hypervisor must have the Xen's own PTI patch"}]
CVE-2017-17566 is un-fixed on Debian jessie
ping @maniackcrudelis @sbadia
Do you know if it's at the LXC level or at the hypervisor level? I remember that gitoyen had updated the hypervisor
The 2 LXC are supposed to be updated everyday. And there's no upgrade available. https://demo.yunohost.org/yunohost/admin/#/update
I know you need to reboot to load the new kernel, is it done/needed here or are they updated before the boot?
Here, the containers are started, then upgraded, and finally shutdown. So a reboot should not be needed here.
Are the containers that are restarted are the one that were shutdown or are they rebuilt every time?
After been shutdown, if there was an upgrade, the container will have a snapshot and then started again when used.
Hi,
Just more infos on this subject, the hypervisor is a Debian stable host (stretch) running the latest xen (with comet enabled) 4.8.3+xsa262+shim4.10.0+comet3-1+deb9u7
and kernel 4.9.88-1+deb9u1
.
The yunohost vm is a Debian oldstable (jessie) running on the kernel 3.16+63+deb8u2
.
Xen side, this vm use pvh + shim as recommended by Xen faq
# /etc/xen/yunohost.srv.gitoyen.net.cfg
kernel = "/usr/lib/grub-xen/grub-x86_64-xen.bin"
extra = "(hd0)/boot/grub/grub.conf"
# https://xenbits.xen.org/xsa/xsa254/README.comet
type = "pvh"
pvshim = 1
As far I understand the check script, it only check if pti is enabled, the vm kernel support kpti, but sys not seems enabled…
root@yunohost:~# grep PAGE_TABLE /boot/config-3.16.0-6-amd64
CONFIG_PAGE_TABLE_ISOLATION=y
root@yunohost:~# journalctl -k -b|grep isola
root@yunohost:~# echo $?
1
And the script
root@yunohost:~# bash spectre-meltdown-checker.sh --variant 3 --batch json
[{"NAME":"MELTDOWN","CVE":"CVE-2017-5754","VULNERABLE":true,"INFOS":"Xen PV DomUs are vulnerable and need to be run in HVM, PVHVM, PVH mode, or the Xen hypervisor must have the Xen's own PTI patch"}]
I missed something ?
root@yunohost:~# cat /proc/cmdline
root=UUID=e70212d0-2804-4a06-8042-293d98d5577b ro pti=on quiet
root@yunohost:~# journalctl -k -b|grep isola
root@yunohost:~#
:-/
Tested, apparently it's fixed now ? I connected to the admin interface and didnt get the warning
Original Redmine Issue: 1079
Author Name: alexAubin
Some users noticed that, on the admin interface of demo.yunohost.org, the spooky message about the server being vulnerable to Meltdown is displayed