Hello, my name is fabian simmer and I'll walk you through the memory team's plans for 14.3. There are three major items: one is or two of them are follow-ups or carry-overs from 14.2. The first one is shifting sidekick workers to database replicas. This is enabled by default in 14.2, but there's a little bit of cleanup to be done. As a reminder, we can actually execute some of our sidekick jobs on read-only database replicas, rather than the primary because they only need read access.

This is very beneficial because the primary database in the postgres or database cluster will have reduced load, and so we benefit from that when you actually use a post cluster for large installations, such as gitlab.com. We're almost done with this. As I said, this is already enabled um by default in 14.2, but there are a few um sort of cleanups that and improvements that we can make in 14.3 that will carry over and we'll be able to close this one out.

The next one is updating to uh ruby version 3.0, so there's a major release of of ruby. We are on movie 2.7 at the moment and we're a bit ahead of the the curve here. We're trying to um get this done before ruby 2.7 is end of life as a major release. We expect some breaking changes, but also some big improvements, and so right now we're focusing on fixing known deprecation warnings and dependencies and making sure that the upgrade goes smooth without any existing functionality. Breaking that is continuing.

We're almost done with this first part of the work, and then we have to roll this out and ensure that all of it works well and behaves as expected.

There's also a second piece here: um we are going to upgrade to ruby 2.74 before we upgrade to ruby three, because there are some interesting fixes in that minor release, and given that the memory team has experience in doing this, we'll pick that up and upgrade to that minor release which is going to be risk. There will be three upgrade even further, because some of the changes to the sort of latest 2.7 release will make it easier to upgrade and that's what we're going to do uh as a stretch goal.

We are going to potentially get to um actually improving some of our sub kick metrics, so there is some work to be done to improve how we emit metrics- and this is specifically relevant when you have a cluster of different sidekick nodes, because they um they have some challenges with how they actually handle prometheus metrics, and this is this is something that we may get to if all of these other things um get done a little bit earlier, but it's a stretch.

So that's 14.3. Thanks for listening and I'll talk to you next time cheers.