NetBox maintainers meeting minutes, 29 Feb 2024

[jeffgdotorg]

NetBox maintainers meeting minutes, 29 Feb 2024

[Jeremy] Bug blitz results

• 2 weeks ago we tagged a few bugs to work on, and we did well against them. Thanks to everybody for the good work.
• Jeff: Can we keep that momentum going by making this exercise a repeating one?
• ↳ There are lots of ways we could change this up to keep it "fun" and impactful. Jeremy and Arthur are covered up with 4.0 work right now, and Dan was not present for this part of the conversation, so let's defer.
• ↳ Decision: Jeff to revisit in next week's call

[Jeremy] FR #15277

• Would like to tag this for v4.0, but sensitive to breaking changes it would effect
• Dan: I don't see a problem. How many users would touch that field in an automation? Wouldn't most work directyl with the generic foreign-key?
• Jeremy: Mostly, but custom fields, custom links, event rule, export template, and saved filter are exceptions. Those types are probably not being utilized very much so maybe we can forge ahead.
• Jeremy: I like this, we'll of course document every breaking change, but it also rolls up nicely into a concise statement: "Any place you see content type today, you should be using object type going forward. Do a global search and replace on your scripts." We'll see how the community reacts.
• Decision: Tag for v4.0

[Jeff] Slightly off-topic but relevant: Floorplan plugin maintainership mess (discussion #15298)

• Jeff: I recently saw Tony active in NetDev slack. I'll approach him about transferring his repo to the netbox-community org. In parallel I'll reach out to the NetBox Labs folks who have been keeping this plugin working with recent NetBox releases (for the benefit of NetBox Cloud customers)

[Abhi] #5858 will this be added in v4.0?

• Jeremy: It's been a very long time since I did any work on this, and if I had a local clone with a branch, it's long gone by now. We've substantially redone the UI since anyway, so we would have to redo the work to display that modal.

[Jeremy] v4.0 milestones update

• #9856: Arthur's work continues on replacing `graphene-django`` with Strawberry.
• ↳ Arthur: Two main items left to work on. Estimate about a week left to complete.
• #12795: Jeremy wanted to take a stab at custom group model. Thinks it's now doing what it should. There's been some PGSQL sequence and index cleanup work implied by this, been handling with just raw SQL renames. PR needs review.
• #15292: Good to go, just needs review
• #14736: Hoping to have a PR ready for review tomorrow; needs lots of testing to be sure we didn't regress anywhere.
• #15043: ETA scooted out slightly so we don't have to go back and clean this up while we're still working hard and heavy on the thing we're documenting.
• #10587: Last major item under scripts-to-DB theme.
• ↳ Arthur: I'm picking that one up tomorrow.
• #15278: This works, but ran into a nasty performance issue. The API test for DCIM takes 39 sec (up from 16 sec) with the changes in place. Working to track down the penalty. If it's something intrinsic to Django serialization code paths, we might have to ditch this and kick down the road for future consideration.
• ↳ Dan: Could this be related to any prefetching you're doing? I've seen similar issues that turned out to stem from that. Jeremy doesn't think so, anecdotal testing isn't consistent with that, but it could be.
• ↳ Jeremy to do more digging. Just be aware one might be at risk for 4.0.
• ↳ We've talked previously about introducing automated performance testing; this challenge illustrates the potential utility of going down that road.
• #15236: Work continues. Please try to make individual issues rather than using this issue as a dumping ground for any breakage you identify going forward. Hope to close this one out pretty soon.
• #9583: Dan's PR needs review.
• #14739: Would still like to tackle. We still should be able to knock it out for 4.0.
• #15153: Strictly window dressing, nice to have.
• #15154: Okay moving, but we don't necessarily have to tie a move to 4.0. The worker we choose is immaterial to our code base.

Dan: PR #15301 (fixes #13722) needs review

• Jeremy will try to take a look.