v1.1? IPv6? about libreqos

[syadnom]

ipv6 is basically a requirement for me. I'm not ready to commit (because I'm still on preseem and haven't made a hard decision to leave them) but if v1.1 could have ipv6 and the stats screen you showed in the other thread, I'd pony up a bit of cash.

[lakelinx]

@rchac Id be happy to up the bounty another $1k if that helps to get it implemented.

[rchac]

@syadnom Added IPv6 support for v1.2 (xdp off) and IPv6 support for UISP integration via MikroTik MAC association. https://github.com/rchac/LibreQoS/tree/main/v1.2

UPDATE: I have tested it in production with success. It works correctly with IPv6, and UISP integration is now working well.

[rchac]

See #98

[rchac]

I appreciate your feedback. I think you're right, I need to engage folks there again. I am still happy to put up the cash reward thing to make it happen. I am hoping with the more flexible hierarchy for HTBs in v1.1, the TCP latency tracking, and (soon?) IPv6 support - that LibreQoS can be a better fit for your ISP environment. I'm already running v1.1 in production and it's pretty nifty.

[rchac]

Idea: I can adapt v1.2-alpha to have an XDP_multicore on/off toggle boolean. If False, it could use a regular one-CPU-core HTB structure with parent nodes within that, allowing IPv6 to work with regular hashed TC filters. It should handle somewhere between 4-6Gbps. Would this be helpful to those who need IPv6?

[syadnom]

Idea: I can adapt v1.2-alpha to have an XDP_multicore on/off toggle boolean. If False, it could use a regular one-CPU-core HTB structure with parent nodes within that, allowing IPv6 to work with regular hashed TC filters. It should handle somewhere between 4-6Gbps. Would this be helpful to those who need IPv6?

This would be interesting. Also would allow 'scale out' with cheaper boxes, ie fast dual and quad cores vs trying to shove tons of fast cores in one box.

[rchac]

Ok great. I'll start experimenting with it.