I recently upgraded my network from a basic router to a Netgear GS108E Gigabit switch. Backups are now way faster, which is great… but now I’ve run into a new problem.
I run Proxmox Backup Server (PBS) in a VM on one of my servers (let’s call it Proxmox 2). PBS has 100 GB of cache storage since I’m using S3 as a datastore - it stores chunks locally before uploading to S3.
Here’s the issue: I try to backup a VM from another server (Proxmox 1). The VM’s storage is 300 GB, but actual usage is under 30 GB. Everything works fine until it hits 34% (104 GB of data). At that point, the PBS VM and the host server itself become completely unreachable. I have to force power off and restart.
I understand why PBS might crash – my VM has 4 GB RAM, 2 CPU cores, and my internet is only around 60–70 Mbps—but why would the host server hang as well?
Has anyone experienced something like this? Could it be networking, storage, or something else in Proxmox causing the host to become unresponsive when PBS gets busy? Any ideas or advice would be appreciated!
For further context: https://lemmy.buddyverse.net/post/5455
You must log in or # to comment.
I recently learned that during a snapshot backup, all subsequent writes are done on the VM and the Backup server until the backup is finished. So if you have the slightliest performance problems with your backup target, you’ll have increasing I/O delays which can result in timeouts and intimately kernel crashes on the host.
You need logs, but I’m certain your OOMKilling the host. Pay less attention to what the VM is doing, and more to what the host is doing. Passing that amount of data through a VM on a host with only 4GB of RAM sounds like it’s destined to fail. Put a hard limit on the memory the VM can use and see if that helps, but I just don’t think you have the resources to manage this in the way you’re attempting.
A better question is: if you’re only running PBS on this machine, why is it in a VM?
