 Welcome to this month's, hello cat. You get to say hi, and then you get to go away now. Welcome to this month's Prometheus team functional update. Let me start the screen share. Everything look good? All right. So welcome, I'm Ben Kochi, I'm the Prometheus team lead. Joshua Lambert is the product manager, Powell, and Kevin and Julius are all engineers on the team. So getting right into it, 9.4 is a great release for us. We have now additional metrics in 9.4. So what this means is users can now add additional metrics to the display page for all of their CI jobs. And this allows for monitoring other common systems and other service metrics. This is super nice because it now allows us to not be limited by simple things like CPU and memory metrics. And we can now start getting into actual business metrics like request handling and latency and things like that. Well, we're also going to be adding, we've also already added support for engine exit and other load balancers for Kubernetes, which is super great. So you can get application metrics and load balancer metrics directly from Kubernetes ingress controllers. We also have been working on instrumenting the Rails app itself. We now have histogram metrics for web request, which is super, super useful. This will be fantastically useful for monitoring the latency of production at GitLab.com. But also users who run large installations of GitLab will also be able to get this useful data out of Unicorn. We also have moved the metrics access from a token that was in the database to a simple whitelist. This allows for much easier configuration, which means we can now include Unicorn metrics by default in the omnibus package Prometheus server, making it much easier for users to be able to monitor their GitLab installs. Coming up at nine five, we're going to be now adding the ability to display multiple time series in a single chart. This will be super useful for looking at both, for example, looking at the 95th percentile versus the median latency on under quest. We're also going to be working on for enterprise edition. We'll be monitoring canaries in production since canary features are enterprise edition feature. So you'll be able to get canary graphs into your projects. We're also continuing to support Prometheus in GitLab.com production using the MTail utility. We've been able to instrument a bunch of things quickly and easily. We now have Wally as the backup utility that we use. We now have production backup metrics and the last status of all the backups. We had some problems with Unicorn worker killer and timeout problems. So we instrumented using the Unicorn logs for that. We also were able to put the sidekick metrics we're also able to put the sidekick metrics back into production and we'll be making improvements to sidekick metrics that we can get much better information about the running sidekick jobs. In FlexDB, the Prometheus bridging is still a work in progress, but that's moving along slowly. Another fun one today is we can now use one of the best things that metrics can provide is it can provide a way to make decisions about running applications. For example, we needed to know what would be a safe threshold to adjust database query timeouts. Instead of just guessing, we were able to go through and process and generate metrics using MTail to find out that, oh, we can adjust timeout down to 15 seconds and only affect a very small number of requests versus the current default of 60 seconds. So you can always use metrics to make decisions and it's much easier to make decisions when you have numbers to back up what you've been working on. And that's pretty much it. Thanks. Does anybody have questions? Are you sharing the slides? Oh, did the slides not show up? I mean, I saw them, but like, my bad. Yes, the URL, I'll update the calendar and invite with the URL. Cool. Thanks, man. Yep. All right. Well, that was easy. Have a good day, everybody.