[SOLVED] Web UI extremely slow after update, but server resources are fine (High "Blocked" time in API calls)

S

sagara1919

New Member
Jul 30, 2025
3
0
1
Hello,

I'm experiencing a severe performance issue with the Proxmox Web UI after recent updates. Any action in the UI, like clicking on a node or a VM, takes about 10 seconds to respond. The server itself is responsive via SSH, and the VMs are running without any issues.

System Information:

  • Proxmox VE Version: Updated to pve-manager 8.4.5
  • Kernel Version: Running 6.8.12-13-pve
  • CPU: AMD Ryzen 7 7735U
  • Memory: 32GB
Problem History:The issue started after performing apt full-upgrade around July 19th to July 27th, where packages like pve-manager (to 8.4.5) and the kernel were updated.

Troubleshooting Steps I've already taken:I've tried almost everything I can think of, but nothing has solved the problem:

  1. Checked Server Resources: top, iostat, etc., show the server is mostly idle. CPU usage is very low, there is no I/O wait, and plenty of free memory.
  2. Stopped All VMs: Shutting down all running VMs does not improve the UI responsiveness.
  3. Rebooted Multiple Times: The issue persists after every reboot.
  4. Browser Issues: Tried different browsers (Chrome, Firefox), cleared cache, and used incognito mode. No change.
  5. Checked PVE Services: systemctl status pveproxy pvedaemon and journalctl -u pveproxy show no errors.
  6. Removed network-manager: I found that network-manager was installed as a dependency. I have completely purged it (apt purge network-manager) and rebooted, but the problem remains.
  7. Booted with Older Kernel: Booted into the previous kernel (6.8.12-12-pve), but the UI was still slow.
  8. Reinstalled pve-manager: Tried apt-get install --reinstall pve-manager, but it didn't help.
  9. Checked dmesg: No critical hardware errors. There was minor link flapping on boot, but the links are stable during operation.
  10. Checked DNS: DNS resolution is fast (dig query time is ~45ms to 8.8.8.8).
  11. Disabled NIC Offloading: Tried turning off TSO, GSO, and GRO via ethtool. No effect.
Key Finding - HAR File Analysis:This seems to be the most important clue. I captured a HAR file while loading the UI.

  • The API call to /api2/json/cluster/resources takes about 9.5 seconds to complete.
  • Looking at the timing breakdown, almost the entire duration is spent in the "Blocked" state. This suggests the browser is waiting for a long time just to establish a connection to the pveproxy service, and that the server-side processing is not the bottleneck.
Given all this, I suspect there might be an issue with how pveproxy is handling incoming connections after the recent updates. Has anyone experienced a similar issue? Any advice on what to check next would be greatly appreciated.

Thank you.
 
Hello Dietmar,

Thank you for your suggestion.

I checked the storage status, but everything appears to be active and online. So, an offline storage causing a timeout doesn't seem to be the issue here.

Here is the output of pvesm status:

# pvesm status　Name Type Status Total Used Available
local diractive 98497780 33018496 60429736 33.52%
local-lvm lvmthin active 832888832 71545150 761343681 8.59%

I've also tried almost everything else I can think of (checking server resources, stopping all VMs, booting with an older kernel, reinstalling pve-manager, removing network-manager, disabling PCIe ASPM, etc.), but the problem persists.

The most significant clue I have is from a HAR file, which shows that API calls like /api2/json/cluster/resources are taking nearly 10 seconds, with almost all of that time spent in a "Blocked" state before the request is even sent.

Do you have any other ideas what could cause this kind of connection blocking on the pveproxy service?

Thank you for your help.
 
Hello Dietmar,

Yes, there is a firewall (a FortiGate) between my client and the Proxmox server.

My client PC is on my local LAN, and the Proxmox host is located in a separate server network segment, behind the firewall. I have the necessary routes and policies in place to allow access from the LAN to the Proxmox management port.

This exact network configuration has been in place for a long time and has always worked perfectly without any performance issues.

The extreme slowness I'm experiencing started immediately after I performed the recent Proxmox updates. I have also checked the firewall logs, and I can confirm there are no denies, IPS alerts, or any other security events that would explain this 10-second delay on API calls.

Is it possible that a change in the newer version of pveproxy is causing an unexpected interaction with the firewall?

Thank you for your continued help.
 
You must log in or register to reply here.
Top Bottom