https://github.com/vyos/vyos-1x/pull/4353

Both kernel and user-space support are needed. On the kernel side, you must enable vDPA-related options (e.g., CONFIG_VDPA and/or CONFIG_VHOST_VDPA) to provide the necessary infrastructure. On the user-space side, binaries that interact with VirtIO devices must be updated to take advantage of this support. For example, QEMU needs to be built with vDPA support so that virtual machines can offload packet processing to the hardware accelerator. Additionally, tools like libvirt or other management utilities might require adjustments to properly configure and utilize vDPA-enabled devices.

Is there any kernel option that should be enabled, or just some binaries or both of them?

PR https://github.com/vyos/vyos-1x/pull/4348

hsflowd expects the agent-address to be chosen from addresses it can see in it's own namespace.

https://github.com/vyos/vyos-build/pull/906

https://github.com/vyos/vyos-1x/pull/4347

https://github.com/intel/ethernet-linux-i40e/issues/12

IXGBE driver version 6.0.5 now builds on Kernel 6.6.77 https://github.com/intel/ethernet-linux-ixgbe/releases/tag/v6.0.5

https://github.com/vyos/vyos-1x/pull/4346

Backport to 1.5-stream-2025-Q2, after Q1 is released.

Can someone provide me with information of why zebra_irdp.so or irdp.so is missing the modules for FRR 10.2.1
I have monitored the starting of FRR protocols and it starts every needed protocols and once it comes to zebra_irdp.so and could ldopen the module zebra fails

I found out that the frr 10.2.1 may be missing some file. As you can see from the snippet below the shared module irdp.so appears to be missing from the package. Immediately after that zebra fails

"Failed to listen on syslog.socket" doesnt look good if you ask me ;-)

The individual tests look good - must be a permutation

It looks like this commit wasnt properly tested before pushed to the public tree according to:

https://github.com/vyos/vyos-1x/pull/4340

@HollyGurza the vyos-network-event logger service could be used to superseede netplugd which we wanted to remove for a long time.

this issue doesn't reproduce on:

The problem still exists. I tried to redirect ipv6 multicast to nflog group via nft. After that, I see multicast traffic but it is encapsulated in gre ip, not in gre ipv6. Maybe this is the issue.

It will require a migration script, but otherwise it's simple to implement a check.

I suppose we could just add a "guesstimate" check, say, 2GB of free space. At the moment, rolling release images are about 600M, 2GB should be enough for a while, hopefully.

The official recommendation will be to build new images. A blog post on the flavor system is coming, and it will be in the docs, too.