On Thu, Jul 23, 2026 at 6:18 PM Jimmy <[email protected]> wrote:
>
> > Which is not how our development process works.  We rely on maintainers
> > pushing code directly to branches where it belongs.  And we trust
> > our maintainers to be careful with that.
>
> It does seem like too big a request to change your development process.
> Yes, the premise of branch protection is moot if the developers still have 
> the same access.

Thanks Jimmy for your comments.

Only the maintainers have the right to create branches and to push the commits.

>
> The goal would be to prevent a compromised or neglectful individual from 
> making direct changes to your branches.
> So yes, it would require a PR with at least one other approval.
>
> There's the additional benefit of ensuring status checks (e.g., tests) are 
> passing before commit.
> This also avoids developers introducing competing commits that work fine in 
> isolation but break each other's commits.

When a developer submits a patch to the ovs-dev mailing list, the
ovs-robot runs the CI tests and the maintainers
ensure that the patch passes all the CI tests before committing.
Patchworks for both OVN and OVS can be found here:

https://patchwork.ozlabs.org/project/openvswitch/list/?
https://patchwork.ozlabs.org/project/ovn/list/?


Thanks
Numan

Numan

>
> Again, I don't want you to have to rethink your entire SDLC, even if I worry 
> about a popular tool lacking the controls I'm familiar with.
>
> Thanks
> Jimmy
>
> On Thu, Jul 23, 2026 at 2:16 PM Ilya Maximets <[email protected]> wrote:
>>
>> On 7/23/26 10:56 PM, Numan Siddique wrote:
>> > On Thu, Jul 23, 2026 at 4:51 PM Ilya Maximets <[email protected]> wrote:
>> >>
>> >> On 7/23/26 10:08 PM, Numan Siddique wrote:
>> >>> On Thu, Jul 23, 2026 at 3:59 PM Ilya Maximets <[email protected]> wrote:
>> >>>>
>> >>>> On 7/23/26 9:52 PM, Numan Siddique wrote:
>> >>>>> Hi OVN and OVS maintainers
>> >>>>>
>> >>>>> Jimmy (CC'ed) has raised this PR issue for OVN -
>> >>>>> https://github.com/ovn-org/ovn/issues/317
>> >>>>> about protecting the release branches.
>> >>>>>
>> >>>>> Are there any reservations against this request ? Or any downsides ?
>> >>>>>
>> >>>>> We are setting up pull mirror rules for our downstream OVS and OVN
>> >>>>> repos and we want to base it on the protected branches.
>> >>>>>
>> >>>>> Looks like OVS branches were protected until branch-3.2.
>> >>>>>
>> >>>>> Can we branch protections for both OVS and OVN ?
>> >>>>
>> >>>> Branch protection is a mechanism to prevent direct pushes, AFAIR.
>> >>>> So, the only way to merge changes becomes a github PR.  Which is
>> >>>> not how our development process works.  We rely on maintainers
>> >>>> pushing code directly to branches where it belongs.  And we trust
>> >>>> our maintainers to be careful with that.
>> >>>>
>> >>>
>> >>> Thanks Ilya for the reply.
>> >>>
>> >>> I think it is possible to have direct pushes from the maintainers even
>> >>> if the branch is protected
>> >>> by adding rules to allow maintainers.  But it won't be possible for
>> >>> the force pushes.
>> >>>
>> >>> Let me know if it makes sense to protect the branches and also allow 
>> >>> maintainers
>> >>> to push.
>> >>
>> >> Maintainers are the only ones with the write access, i.e., the only
>> >> ones who can push.  Is there a point in protection rules if everyone
>> >> who can push will still be able to do so?
>> >>
>> >> We do also allow maintainers to use force-push for quickly fixing
>> >> their mistakes.  And mistakes are a part of having write access.
>> >> Obviously, it's not something that should be used lightly or in any
>> >> way frequently.  But it's good to have a tool when it is necessary.
>> >
>> > Got it.  I agree there and I myself have done a few mistakes before.
>> >
>> >>
>> >> I'm also not sure what is the original idea behind only mirroring
>> >> protected branches?  Could you elaborate?
>> >>
>> >
>> > From what I understand, security wants to only mirror the upstream
>> > code from the branches which are protected.
>>
>> Fun fact:  I just checked and it is possible to create a branch protection
>> rule with the following configuration:
>>
>>  1. Allow force pushes (for everyone with push access)
>>  2. Allow branch deletions (for everyone with push access)
>>  3. Require linear history.
>>
>> This rule adds no real restrictions that are meaningful to our development
>> process.  The only restriction is a linear history that we use anyway.
>>
>> But, after applying it to a branch, it now shows a shield icon with the
>> 'This branch is protected with branch protections' legend.  And since it
>> is a legacy branch protection rule, nobody except for the owner can see
>> what this protection rule actually is.
>>
>> Similar thing can be done with the branch 'ruleset', but people can see
>> what the ruleset is enforcing byt clicking on it.
>>
>> Though rulesets allow creation of a fully restrictive rule and make it
>> bypassed by everyone with the Write role.  Bypasses are not reported in
>> the UI, so there will be a rule, but it will never be enforced in practice,
>> while being reported as 'Active - This ruleset will be enforced'.
>>
>> All in all, unless you're the organization owner, the branch protection
>> status on the branches page means absolutely nothing.
>>
>> >
>> > Maybe @[email protected]  can perhaps elaborate more as he
>> > is the one who created the ticket.
>> >
>> > Thanks
>> > Numan
>> >
>> >>>
>> >>>
>> >>>> Branch protection is enabled on OVS branches that are no longer
>> >>>> maintained to prevent accidental pushes.  They are just frozen in
>> >>>> time.  Same could be done for OVN.
>> >>>
>> >>> I see.  This explains why older branches have protection.
>> >>>
>> >>> Thanks
>> >>> Numan
>> >>>
>> >>>
>> >>>  But we can't do that for
>> >>>> currently maintained branches.
>> >>>
>> >>>>
>> >>>> Best regards, Ilya Maximets.
>> >>
>>
_______________________________________________
dev mailing list
[email protected]
https://mail.openvswitch.org/mailman/listinfo/ovs-dev

Reply via email to