Network behavior with Hardened Repo ISO

Hi ​@gravelrider 

Hi, in my opinion there is no problem with the ISO. The problem is that the UI interface is fragile and unreliable during LACP configuration.
LACP is stateful and timing-sensitive amd slight differences cause very different outcomes.
Here are some possible causes of the abnormal behaviour; NIC enumeration order changed,
one NIC negotiated faster, switch port already had cached LACP state, timing difference during boot.
The recommended advice is always to skip network configuration in the UI entirely and configure everything via NMCLI post-install.
Manual bonding is the real-world best practice.
regards   

Hi ​@gravelrider -

I honestly haven’t played with the VHR ISO much, but generally...when I configure any networking within Linux, I do so as ​@Link State suggested (after install). I recommend trying that alternative route and see if it works. At the very least, you can ping Veeam Support and see if they’re able to replicate your configuratin behavior and see if there is a bug.

Keep us posted.

Best.

I have use VHR ISO with LACP with some environments and everything is working fine.

Have you tried to configure the LACP on avanced network mode?

This mode is more complete.

Thank you all. Yes, manual setup works fine. No biggie.
Good to hear others have seen issues with the UI.

Hi,

is that advise still in place to used cli, when it comes to the Infrastructure ISO V13. There is only the nmtui interface (with advanced) with also strange behaviour. IMHO there is no (easy) way to get into CLI to configure it. Here it is a config with a tagged vlan on a LACP bond.

Thanking in advance,

Martin

​@MartinEW - your best bet is to contact Veeam Support to get official direction.

Best.

The primary limitation observed with LACP and link aggregation is not a Veeam-related issue, but rather an inherent characteristic of the IEEE 802.1AX (LACP) networking standard.

This behavior is driven by how TCP traffic is distributed across aggregated links. By design, a single TCP session is mapped to a single physical interface within the link aggregation group. The hashing algorithm used by LACP (typically based on source/destination IPs and ports) ensures packet order is preserved by keeping each TCP flow on the same member port.

Conceptually, LACP does not increase the bandwidth of an individual TCP stream. Instead, it allows multiple concurrent TCP streams to be distributed across the aggregated links, enabling load sharing at the flow level rather than per-packet. This design prevents packet reordering while providing aggregate throughput scaling when sufficient parallel flows exist.

The Hardened Repository ISO installer configures networking using an early-stage NetworkManager-based workflow. When LACP (802.1AX) is selected during installation, the installer expects the bond to transition to an active state immediately. However, LACP negotiation is asynchronous and dependent on the switch fully establishing the aggregation. If all member interfaces do not reach an operational state fast enough, the installer UI can remain stuck in connecting and eventually time out, resulting in a disconnect.

This explains several of the symptoms you’re seeing:

• Interfaces are detected but fail to connect during the installer workflow
• The switch reports links as up, but the installer never sees a completed bond
• A single NIC configured temporarily works, allowing system access
• Manual bonding via CLI works consistently after install