My company runs a handful of OpenVZ servers (HN) and hosts all the server as containers on these HNs.
We have done some pentest and found out we have to do update the kernel on the HNs to resolve a potential security issue. The servers have huge amounts of storage and are geographically scattered so having a redundant server for every single server is out of the question. But the HN themselves aren't so big. So here's the scenario I have come up with.
- Back up the HN operating system which is CentOS (different versions from 5.4 to 5.6)
- Perfrom the yum update
- Reboot the HN but skip fsck on any partition other than the HN OS partition
- Unmount the containers one by one and fsck them one by one
Can you think of a better scenario? Another thing I thought of was somehow replicating the host machine and trying the update on that replica before actually running it on the real machine.
Any idea how I can do that? This is my machine. I only need the
/ for doing the host upgrade.
[root@host ~]# df -h Filesystem Size Used Avail Use% Mounted on /dev/sda3 1.8T 13G 1.7T 1% / /dev/sda1 99M 46M 48M 50% /boot tmpfs 16G 0 16G 0% /dev/shm /dev/sdb1 1.8T 350G 1.4T 21% /mnt/disk0 /dev/sdb2 1.8T 208G 1.5T 13% /mnt/disk1