M #: auto virtio queues for network interfaces - #258
Conversation
|
@dann1 Hi, I think #243 is something we will move forward with eventually, the whole role is havily refactored there, so we have structures (and 'variants') that we can merge instead of templates. Even if the cgroups solution for CPU isolation present there is not accepted we should still go with it, I believe. Please give it a look, as I said if you don't want the cgroup stuff we can extract just the basic variants. 🤔 |
|
I mean if we don't, then at some point down the line that xml template will become so unreadable it's going to be a nightmare to read and modify (it already is tbh). 🤷 |
|
Ah perfect, I'll take a look at it in there. Sorry I haven't had time to review things. Just putting out fires and trying move things forward. Reviews are not ignored 🤗 Should I add this commit (or its equivalent) to that branch then ? I see that we still have the 2 queues in dpdk only. |
|
@dann1 Please take a look first if you can to understand the madness there 🥹 , I mean what is the proposal actually. If you must, we can merge this one no problem, I'm just saying maybe it's worth to refactor it first :). It's up to you and your review anyway. 🤔 |
Set frontend VM nic queues automatically based on VCPU, following opennebula behavior.