VE: Check Package for Host, Guests, Storage, and Clusters

If the agent is installed on a Proxmox host, octoja already displays the VMs and containers, storage, snapshots, and corresponding actions on the device page. What’s missing is a monitoring feature that constantly keeps an eye on the node and alerts you if something goes wrong. Such a feature already exists for VMware and Hyper-V.

The following scenarios would be useful:

• A VM or container isn’t running even though it should be.

• A guest is set to "start on boot" but has stopped responding.

• A guest is stuck in a lock because a backup or migration was aborted.

• A storage device is running out of space or is no longer accessible, such as a mounted NFS or CIFS share.

• A cluster node is offline or the quorum is missing. On a single node without a cluster, this item should simply be omitted without an error message.

• Snapshots are getting too old or there are too many of them. They take up storage space and slow down the VM, and they’re rarely cleaned up.

• Data backup: when was the last successful backup for each guest, and has a job failed recently? This can be addressed later; the rest is more urgent.

Ideally, this should be a single check where the individual items can be disabled, with custom thresholds for storage usage and snapshot age, and an exception list for test and template guests. Login credentials shouldn’t be necessary, since the agent runs on the node. And if the inventory detects that a device is a Proxmox host, the check could be suggested right away, instead of having to set it up manually on every node.

There’s no need to check what’s already covered by existing checks on the host: ZFS pools, disk health and RAID, CPU, RAM and disk space, services, and the web interface certificate.