In my last job, we always created snapshots before we did any system or package upgrades. This got me thinking: are snapshots really enough? How likely is it that these upgrades could cause persistent metadata corruption on the disk, which wouldn't be fixed by a snapshot? Should backups be used in addition to snapshots for maintenance procedures?
4 Answers
Snapshots are awesome for quick rollbacks. They let you go back to how the disk was before the update. But remember, they only work within the VM context. If something else is changing data outside of the VM, like updates on a remote database, reverting a snapshot won't fix those changes. In terms of metadata corruption, the risks are minimal with solid virtualization, but you still need backups as a safety net. Mistakes can happen, like reverting to the wrong snapshot!
While snapshots can help with quick recovery, they are on the same disk as your data. If that disk fails or gets corrupted, you lose both your snapshots and backups. I think for routine updates, snapshots work just fine, but you should always combine them with reliable backups just to be safe.
It really comes down to your situation. Are your machines like pets or cattle? Do you have compliance needs for backups? Personally, I'm okay with taking down some Kubernetes nodes without backups since our data isn't held long-term on disk. But it’s a case-by-case situation.
It really depends on the type of snapshot. For instance, a VM snapshot taken from the hypervisor isn’t influenced by changes in the VM's OS, which means it can’t be corrupted by software changes. On the other hand, file system snapshots could potentially have issues since they are controlled by the file system driver connected to the kernel. Plus, if you revert to a snapshot, you're restoring everything from that moment, whereas a backup allows for selective restores.
Related Questions
Can't Load PhpMyadmin On After Server Update
Redirect www to non-www in Apache Conf
How To Check If Your SSL Cert Is SHA 1
Windows TrackPad Gestures