[GH-ISSUE #39] RollingUpdate connects to unrouted IPs #29

New issue

Closed

opened 2026-03-02 15:47:21 +03:00 by kerem · 2 comments

kerem commented

2026-03-02 15:47:21 +03:00

Owner

Originally created by @sundown78 on GitHub (Feb 5, 2026).
Original GitHub issue: https://github.com/PegaProx/project-pegaprox/issues/39

Originally assigned to: @mkellermann97 on GitHub.

Hello,

We have a 3-node Proxmox cluster running. In addition to the management address (10.50.255.x/23), the hosts have additional network cards with non-routed IPs for e.g. iSCSI (10.10.10.0/24) or NFS (172.16.50.0/24).

PegaProx is installed as a virtual machine within the management network, and adding the cluster via the management address worked without any problems.

When using the rolling update function, PegaProx attempts to reach individual hosts on the non-routed IPs at irregular intervals, which of course fails. The management addresses are accessible at all times. It is therefore not clear when or why attempts are made to connect to the non-routable addresses.

[PegaProx_reker-cluster01] INFO: SSH retry 3/3 to 10.10.10.42... [2026-02-05 07:24:59,869] [PegaProx_reker-cluster01] ERROR: SSH to 10.10.10.42 failed: [Errno 101] Network is unreachable [2026-02-05 07:24:59,869] [PegaProx_reker-cluster01] ERROR: [ERROR] Node update failed for reker-pve03: SSH connection failed to root@10.10.10.42:22 - Check if password auth is enabled on the node or configure an SSH key / SSH Verbindung fehlgeschlagen - Prüfe ob Passwort-Auth auf dem Node aktiviert ist oder konfiguriere einen SSH Key

Is there a configuration option to specify that only the management addresses may be used for SSH connections?

regards
René

Originally created by @sundown78 on GitHub (Feb 5, 2026). Original GitHub issue: https://github.com/PegaProx/project-pegaprox/issues/39 Originally assigned to: @mkellermann97 on GitHub. Hello, We have a 3-node Proxmox cluster running. In addition to the management address (10.50.255.x/23), the hosts have additional network cards with non-routed IPs for e.g. iSCSI (10.10.10.0/24) or NFS (172.16.50.0/24). PegaProx is installed as a virtual machine within the management network, and adding the cluster via the management address worked without any problems. When using the rolling update function, PegaProx attempts to reach individual hosts on the non-routed IPs at irregular intervals, which of course fails. The management addresses are accessible at all times. It is therefore not clear when or why attempts are made to connect to the non-routable addresses. `[PegaProx_reker-cluster01] INFO: SSH retry 3/3 to 10.10.10.42... [2026-02-05 07:24:59,869] [PegaProx_reker-cluster01] ERROR: SSH to 10.10.10.42 failed: [Errno 101] Network is unreachable [2026-02-05 07:24:59,869] [PegaProx_reker-cluster01] ERROR: [ERROR] Node update failed for reker-pve03: SSH connection failed to root@10.10.10.42:22 - Check if password auth is enabled on the node or configure an SSH key / SSH Verbindung fehlgeschlagen - Prüfe ob Passwort-Auth auf dem Node aktiviert ist oder konfiguriere einen SSH Key` Is there a configuration option to specify that only the management addresses may be used for SSH connections? regards René

kerem

2026-03-02 15:47:21 +03:00

closed this issue
added the
bug
label

kerem commented

2026-03-02 15:47:22 +03:00

Author

Owner

@mkellermann97 commented on GitHub (Feb 5, 2026):

Hello @sundown78 ,

this issue is currently known to us and will be fixed with the next version.

Regards,
Marcus

@mkellermann97 commented on GitHub (Feb 5, 2026): Hello @sundown78 , this issue is currently known to us and will be fixed with the next version. Regards, Marcus

kerem commented

2026-03-02 15:47:22 +03:00

Author

Owner

@mkellermann97 commented on GitHub (Feb 8, 2026):

Hi @sundown78 ,

please update PegaProx to the newest version.

Regards,
Marcus

@mkellermann97 commented on GitHub (Feb 8, 2026): Hi @sundown78 , please update PegaProx to the newest version. Regards, Marcus