ProxCenter v1.4.10 Release
It brings Site Recovery on ZFS with recovery per guest, a richer vDC tenant model, 53 Prometheus metric families with Grafana dashboards, and warm migrations without VMware Tools.
ProxCenter v1.4.10 is a feature release focused on disaster recovery, multi-tenancy, observability and migration robustness. It introduces Site Recovery on ZFS with recovery per guest, a richer vDC tenant model, 53 Prometheus metric families with Grafana dashboards, disk latency monitoring, and warm migrations from sources without VMware Tools.
🛟 Site Recovery
A ZFS storage engine
Site Recovery now supports ZFS alongside Ceph RBD, with test failover by clone and a fixed target node.
Recovery per guest
Emergency DR now acts per guest: the replica state is shown on each row, a replica can be started from a chosen restore point and stopped again, and plan failback moves to the plan header.
Starting a replica claims that one guest instead of pausing the whole job, and a test failover suspends only the guests under test while their siblings keep replicating.
Replication improvements
- DR replicas are named apart from production, by prefix or suffix, editable after creation
- A replication network per connection, an initial sync streamed as a sparse diff, and failback across differently named Ceph pools
- A new interval schedule mode, so a 30 minute RPO no longer replicates every 10 minutes
- A guest already replicated by another job is refused, and an interrupted cleanup keeps its retry button until everything is really cleaned
🏢 Multi-tenancy and vDC
Fine-grained guest rights
The vm.config permission is split into fine-grained rights covering media, NIC link, NIC, hardware and boot, and the guest interface follows the rights a user holds.
Compute, ISO and network policies per vDC
- A compute policy per vDC, unrestricted or limited to chosen CPU models
- Read-only ISO libraries with an optional tenant upload area
- A VXLAN transport per vDC, with operator peers, a dedicated transport network and MTU, and a tenant-wide VNI for a network stretched across clusters
- Provider SDN VNets accepted as shared uplinks, with their allowed VLAN pool in the deploy wizard
SSO mapped to tenants
SSO groups can now map to a role inside a tenant or a single vDC, and signing out also ends the identity provider session. An SSO login no longer re-adds its own role beside the one an administrator set.
📈 Monitoring and alerts
Prometheus and Grafana
The Prometheus exposition grows from 7 to 53 metric families, including guests with no backup at all, and ships with five Grafana dashboards for fleet, nodes, guests, storage and backups.
Guest disk latency
Disk latency is now visible in the inventory and storage columns, in a Disk I/O curve, through a per-storage alert and in a dedicated dashboard widget.
More signals
- Every installed package is scanned for CVEs
- A Ceph HEALTH_ERR names the conditions behind it, Ceph chart times follow the browser timezone, and cluster OSD flags can be set from the page
- An exclude pattern for stale snapshot alerts, and alert emails that carry readable values
🚚 Migration
- A source without VMware Tools is powered off hard at cutover instead of waiting thirty minutes on an operator, and IDE and SATA disks keep their bus
- A warm migration whose source vSphere the node's VDDK cannot read is refused up front, and a failed block apply shows the real error
- Warm migrations no longer wedge NBD devices on Linux and LVM guests
- Parallel NFC disk downloads with a concurrency slider, and accurate thin disk progress
- Cross-cluster prerequisites (HA, replication, snapshots) are lifted from the dialog, then restored or rolled back
- Cutover, cancel, force power off and root disk choice sit on the job row and survive a page reload
⚖️ DRS
A new History tab shows the reason behind each migration and what constrains placement. Balancing now happens inside each placement domain, guests are weighed by their host memory footprint, and migrations the target node cannot run are no longer proposed.
✨ Also in this release
| Area | What's new |
|---|---|
| 🧩 Dashboard | Guests per Node widget with live CPU and RAM; dashboards export and import as one JSON file |
| 🔎 Search | The command palette finds guests by IP, MAC and notes, and nodes by IP |
| 💾 Backups | A PBS datastore attaches to a cluster from the inventory, with a sub-token limited to it |
| 🔌 Hardware | USB and PCI passthrough through datacenter resource mappings |
| 📦 Containers | LXC features editor and rename through the hostname |
| 🖥️ Node shell | Fullscreen and a detachable window |
| 📄 Reports | PDF template Customization tab with live preview; report data downloadable as CSV |
| 🎨 White label | Tenant branding restored for super administrators, with the favicon and tab title fixed |
| 🛡️ Security | Advisories closed on next, sharp, svgo and the AI SDK chain |
Plus nested pools shown as a collapsible tree, a self-updating cloud image catalog, change tracking exported as CSV, and rolling updates no longer blocked by a subscribed node carrying only enterprise repositories.
⬆️ Upgrading
Eight database migrations are included and run automatically at first boot, and the orchestrator updates its own schema when it starts. Nothing has to be edited by hand, on a standalone installation as on an HA cluster.
docker compose pull && docker compose up -d
Prometheus users
proxcenter_backup_age_seconds now carries node, name and type labels. Any aggregation over the bare metric needs a max by (vmid) or an equivalent.
Full instructions are available in the ProxCenter update guide.
Thank you to everyone who contributed feedback, testing and issue reports for this release.