Hello everyone, I'm running into a serious bottleneck with my backup strategy. I have a 10-node Proxmox cluster, and all VMs are backed up to a single Proxmox Backup Server.
The issue is: I created a cluster-wide backup job selecting VMs by their IDs. When the schedule hits, all 10 nodes trigger their backups simultaneously. Even though I set
To optimize this, I have already enabled
I've considered an alternative where I create separate backup jobs for each node (setting Selection Mode to "All") and manually stagger their schedules.
However, with 10+ nodes and VMs frequently migrating between them, managing a dozen different backup schedules is a management nightmare. Is there a more elegant "cluster-wide" way to throttle or limit concurrent backup tasks so the whole cluster only runs a fixed number of backups at once?
The issue is: I created a cluster-wide backup job selecting VMs by their IDs. When the schedule hits, all 10 nodes trigger their backups simultaneously. Even though I set
IO-Workers to 1 in the Advanced settings, it only limits concurrency within a single VM's process; it doesn't stop all 10 nodes from hitting the PBS at once.To optimize this, I have already enabled
fleecing, which successfully reduced the IO delay on the VM side and kept the guests running smoothly. However, the root cause—the PBS backend—is still getting hammered by data from 10 nodes at the same time, leading to extreme load on the backup server.I've considered an alternative where I create separate backup jobs for each node (setting Selection Mode to "All") and manually stagger their schedules.
However, with 10+ nodes and VMs frequently migrating between them, managing a dozen different backup schedules is a management nightmare. Is there a more elegant "cluster-wide" way to throttle or limit concurrent backup tasks so the whole cluster only runs a fixed number of backups at once?
