596 Connection timed out if the API is asked for /qemu otherwise all fine

[layer7.net]

Hi,

we have here a strange behavior of the proxmox API.

pve-manager/9.1.2/9d436f37a0ac4172 (running kernel: 6.17.2-2-pve)

If we ask via API for

/api2/json/nodes

then all is fine and the response comes immediate:

::ffff:192.168.178.20 - root@pam [23/01/2026:15:54:56 +0100] "GET /api2/json/nodes HTTP/1.1" 200 1957


but if we ask for

/api2/json/nodes/n1/qemu

then at first nothing will appear in the logs ( pveproxy/access.log or systemd pveproxy )

then after some time in access.log we will see:

::ffff:192.168.178.20 - root@pam [23/01/2026:15:54:47 +0100] "GET /api2/json/nodes/n1/qemu HTTP/1.1" 596 -

the curl call will show:

* TLSv1.3 (IN), TLS handshake, [no content] (0):
* TLSv1.3 (IN), TLS handshake, Newsession Ticket (4):
* TLSv1.3 (IN), TLS handshake, [no content] (0):
* TLSv1.3 (IN), TLS handshake, Newsession Ticket (4):
* TLSv1.3 (IN), TLS app data, [no content] (0):
< HTTP/1.1 596 Connection timed out
< Cache-Control: max-age=0
< Connection: close
< Date: Fri, 23 Jan 2026 15:00:13 GMT
< Pragma: no-cache
< Server: pve-api-daemon/3.0

The API of node n1 is called and asked for its own /qemu

What could be the reason? Why is the API call one time successful without issue if NOT asked /qemu while another API call from the same client to the same server asking for /qemu will end up in connection timed out?

This happened occasionally. As soon as a 6th node was added, it happens >90% of the time.

If the proxmox UI is used, there are no issues. All nodes can access each other without issues and show the qemu instances.

Any input is appreciated. Thank you!