ProxCenter v1.4.14

A Ceph cockpit, bulk power actions, multi-node maintenance.

Fixed in priority

• Requests authenticated with a valid API token no longer hang: since 1.4.13 they looped on the license check until the proxy timed out, and grew the frontend's memory until it restarted.
• Scheduled reports run: no schedule had ever fired, nor the nightly report retention cleanup. They now run from the leader, once per pass, with no restart after an edit.

New

• Ceph cockpit: health rail, host map, OSD table with latency, IOPS and throughput, pools, CRUSH, services and performance, at about a sixth of the former Proxmox calls (history is Enterprise).
• Power actions on a selection of guests, across nodes and clusters, from any guest list, the tree or the node menu.
• Several nodes of a cluster in maintenance as one drain, so a guest moves once and HA brings it back to the right node.
• Available updates notification: package updates pending on the Proxmox VE nodes, sent through the notification channels and shown as an info alert (Enterprise, off by default).
• Brand name and logo in every notification email, white label included.
• Inventory > Network lists the guests behind each VNet, bridge and VLAN, across every node of the cluster.
• One bridge and VLAN picker in every NIC, create and migration dialog, searchable by name, tag or zone.
• Running disk moves show their progress in the task panel, with speed and ETA.

Fixes

• A deleted guest no longer comes back after a reload, and creates, clones and migrations refresh one connection only.
• The first console request no longer fails with a 500 while the connection host is down.
• The migrate dialog offers a target storage only for local disks, lists the shared disks that stay put, and refuses a running TPM state move.
• HA is registered on a clone that lands on another node, which therefore starts.
• PBS backups keep their name from the group notes after retention prunes the annotated snapshot.
• VMware adapters of every model keep their source MAC, Flexible, VMXNET3 RDMA and SR-IOV included.
• A fresh install sends plain HTTP to HTTPS, and the login page explains when sign-in needs HTTPS.
• An empty vDC inventory shows the guest list with Create VM and Deploy from template.
• DRS node names stay readable on a crowded per-node chart.
• The change journal ignores the order of OVS options, which logged a change at almost every sweep on OVS bonds.
• Info alerts show as info in the notification bell.

Security and isolation

• IaaS tenants no longer reach the provider's cluster and node routes: they get a 404 before any Proxmox, SSH or database call.
• The AI assistant is off for IaaS tenants, API included.
• A guest's backups are limited to the tenant's PBS namespaces on a PBS shared between tenants.
• Microsegmentation no longer applies an incomplete base group as if it were complete.
• Drains and rolling updates check container storage before any move, so a container on local storage is never moved, and rolling updates now evacuate containers.

Dependencies

jose 6.2.14, @mui/utils 9.5.0, @peculiar/x509 3.1.0, glob 14 under rimraf, npm minor and patch groups, SonarQube scan action 8.4.0.

Upgrading

Pull the images. One schema migration ships with this release, applied by the frontend entrypoint at first boot; the orchestrator creates its new Ceph history tables itself.

• Existing report schedules start sending at their next occurrence after the upgrade. Passes missed by more than an hour are skipped, so nothing fires in a burst.
• Per-OSD I/O in the Ceph cockpit needs ceph-exporter on each node, reachable from the orchestrator on port 9926 (set per connection). The page shows the install steps when it cannot read a node.
• Ceph history is kept 30 days by default, adjustable in Settings > Alert thresholds (scheduler.ceph_metrics_retention_days in config.yaml sets the starting value).
• IaaS tenants that relied on cluster or node screens now get a 404 on them, by design.

Docker Images

docker pull ghcr.io/adminsyspro/proxcenter-frontend:v1.4.14
docker pull ghcr.io/adminsyspro/proxcenter-orchestrator:v1.4.14
docker pull ghcr.io/adminsyspro/proxcenter-weasyprint:v1.4.14

Verify the images

gh attestation verify oci://ghcr.io/adminsyspro/proxcenter-frontend:v1.4.14 --owner adminsyspro

The CycloneDX SBOM of the frontend image is attached below.

Air-gapped bundle

Community: the proxcenter-community-1.4.14.tar.gz asset below. Enterprise: your account on proxcenter.io, Air-gapped tab. Guide: https://docs.proxcenter.io/getting-started/air-gapped-installation