[ovs-dev] OVN Mailing List Suggestions

Mark Michelson mmichels at redhat.com
Tue Sep 1 19:47:56 UTC 2020

Hi all,

For the past eternity, there have been light discussions on various 
mediums[1] about splitting OVN to its own mailing list(s). As it stands, 
OVN discussions and patch submissions are disjoint from OVS, so 
separating it out to its own list makes sense. We[2] did some asking 
around to try to see if anyone that we know to host mailing lists might 
be willing to host an OVN mailing list. Everyone said no. This includes 
the Linux Foundation, who already host the OVS lists.

A few weeks ago, I looked into publicly available mailing list hosts. 
The issue there is that, of course, money is a factor. And essentially, 
you get what you pay for. The cheap options were missing key features 
and didn't offer much support or reliability guarantees, and the 
expensive ones were expensive.

This led to a discussion today on a call between Red Hat and VMWare 
about the topic. We came up with some ideas about how to move forward 
with this.

Option 1: Use Google Groups. Dan Williams and I did a quick and dirty 
test with the ovn-kubernetes Google Group and found that patch emails 
come through properly and can be applied directly to the code. So there 
should be no issue in trying to continue our current workflow with it.

Option 2: Use Github Pull Requests. This would at least get patch 
submissions off the OVS list. We would still need to use the OVS list 
for OVN discussions though (unless we think Github issues are 
appropriate). Github PRs also gives us a simpler way of adding CI.

Option 3: A gracious volunteer offers to host the OVN list(s). If 
someone reading this is willing to have their company host the mailing 
list or host it themselves, then that'd be cool. Understand that this 
would require a good amount of work to maintain.

Option 4: A gracious volunteer offers money to pay for the list(s) to be 
hosted by one of the public companies that hosts mailing lists.

Option 5: Do nothing. If email filters are good enough, then so be it. 
We can drop the topic entirely and never bring it up again.

I'm looking for feedback about these options. Which of these is most 
appealing to you? If one of the options specifically is a problem for 
you or your workflow, why is it a problem?


[1] Though oddly not very often on this list
[2] OVN devs from Red Hat and VMWare, along with folks in management 
from both companies

More information about the dev mailing list