On my 750gr3, I’m seeing some odd behavior with traffic routed between VLANs with hardware offload enabled on bridge ports. Without hardware offload, traffic is routed between VLANs as I’d expect. However, once I enable hardware offload, traffic between random VLAN hosts sporadically fails to flow one moment but flows just fine at a different time. For example, my security cameras are isolated on their own VLAN. With hardware offload enabled, pings (and establishing streams) to one or more cameras will time out for a period of time while pings to other cameras will be normal. At a different time, the camera(s) that timed out before will work fine while another camera that worked fine will time out.
Is it possible hardware offload is bugged on the 750? It looks like switch VLAN configs still aren’t possible with the 750, so it implies MT hasn’t fully implemented full VLAN capabilities of the 750’s switch chip. Granted, it’s not the end of the world. It’s a home network, and offloading camera to NAS traffic (L2 on the same VLAN) is the only real benefit. It still helps cut baseline CPU usage down though, which marginally helps with online gaming latency when internet throughput is maxed.
750Gr3 can HW offload L2 (switching) but can’t HW offload L3 (routing).
It is quite likely you’ve got some configuration error, but we can only guess if you don’t post (complete) config. Things are interleaved and every config line matters (even if it occurs as not important to you).
The only hardware offload 750Gr3 can do is bridge vlan filtering, and that starting with RouterOS 7.1rc5, so what RouterOS version and settings are we talking about?
It is possible that the behavior you are observing with hardware offload enabled on the MikroTik RouterBoard 750GR3 could be due to a bug in the RouterOS software or a problem with the hardware itself. It is also possible that the issue could be caused by a configuration error or some other factor.
To troubleshoot the issue, you could try the following steps:
Check for any RouterOS software updates and apply them if available.
Check the RouterOS log for any error messages or other clues about the cause of the problem.
Try disabling hardware offload and see if the issue persists.
Check the configuration of the VLANs and make sure that they are properly set up and configured.
Check the cables and connections between the routers and devices to ensure that they are secure and functioning properly.
Check the hardware and make sure that it is functioning properly and that there are no issues with the switches or other components.
If you are unable to resolve the issue with these steps, you may want to consider contacting MikroTik support for further assistance. They may be able to provide more specific guidance or troubleshooting steps based on your specific configuration and setup.
Thanks for the replies all. I’ll upload the config tonight, but I may have sorted it. To answer a couple questions though, this is L2 HW offload on ROS 7.6, and the issue isn’t observed with HW offload disabled. Enabling unknown unicast flood appears to have resolved it. Given the randomness, it’s a bit difficult to explicitly test, but I haven’t experienced timeouts since enabling and fairly reliably have them with it disabled.
My guess is that with offload enabled, the destination host isn’t always known on the bridge if a host doesn’t do much L3, and flooding unicast addresses that. I do see “external” hosts in the table (from the switch chip?) including the cameras where I most noticed the issue, so I’m not sure why that would be the case though. I’m going to try to replicate with unicast flooding disabled while watching the hosts table to try to confirm.