text stringlengths 0 1.82k |
|---|
**Carlisia Thompson:** I'm gonna have to end this call and go look up how allocations are done in Go, how to recognize them and how to solve them. I still don't know what you guys are talking about... |
**Erik St. Martin:** So actually there was a talk by Björn Rabenstein recently from the Prometheus team. He did a talk we'll link in the show notes, and I believe some of that he walks through the allocations, and you can actually see the change... So that might be a good look at it. |
**Carlisia Thompson:** Awesome. |
**Erik St. Martin:** And I think there's another one too, I'll look it up for you and we'll put it in the show notes when I find it. I think there was another one that I'd seen a while back too, where somebody was walking through looking at code and determining whether something was gonna be stack allocated or allocate... |
**Brian Ketelsen:** We talked a little bit earlier about the static analysis tools, maybe there's an opportunity for a really simple static analysis tool that helps you understand the allocations that are being made. I know you can use the benchmark tools that are built in to show you allocations, but a nice SSA tool t... |
**Erik St. Martin:** Well some of them get hard too, because if they could suggest the change, you could make the change. There's two tools that are really good for it. One is just using the standard benchmark and test stuff where you can do the pprof, and you can look at it in the pprof tool, or you can visualize it. ... |
**Carlisia Thompson:** Yeah, I can see that. |
**Erik St. Martin:** Like you said, most of the time for most people it doesn't matter. When you're trying to hit the scale... How many requests a second did you say you guys were shooting for, Travis? |
**Travis Reeder:** A million. |
**Erik St. Martin:** Yeah. See, when you wanna hit scales like that, then allocations matter, because a garbage collector comes into play, taking up time, allocations take up time... So when you wanna hit scale like that, or Prometheus... I think they were just talking about half a million requests a second. When you w... |
Speaking of taking on things new, two questions I wanted to ask: one is hiring - for you, Travis, how has that been? We haven't really had to do too much hiring for Go people, so I don't know how big the pool is. Do you guys struggle to find talent? Do you train talent? |
**Travis Reeder:** Yeah, that's a good question. It's easy to find people that want to work in Go, it's harder to find people that have actually worked in it, though. So mostly people we bring on have played with it, but they have good systems engineering backgrounds or things like that. The knowledge you have transfer... |
**Erik St. Martin:** Or you're referencing a Ruby app that you're rewriting in Go? \[laughter\] |
**Travis Reeder:** Those days are pretty much over, but yeah. It's mostly news stuff now. |
**Brian Ketelsen:** Have you had any new hires that came in and picked up Go and just walked aways saying, "Go isn't for me. This isn't the language that I love, I'm gonna go back to that other thing (whatever it is)"? |
**Travis Reeder:** No, that's never happened, actually. Never happened. |
**Brian Ketelsen:** There you go. That's anecdotal proof right there, that Go is awesome. |
**Erik St. Martin:** It's very good, statistical evident. A sample size of one. |
**Carlisia Thompson:** Yeah... You're gonna get a lot of that here. |
**Erik St. Martin:** The other thing that you guys adopted early on too is containers. You kind of spoke to Solomon announcing Docker at the Go SF meetup. |
**Travis Reeder:** Yeah. |
**Erik St. Martin:** How soon after he announced Docker did you guys start kind of playing with it and releasing it? |
**Travis Reeder:** I forget the actual timeline, but I don't think it was much more than a year probably, after he first launched it. It was pretty funny when he launched it that the day he launched it, he pushed it to GitHub that day, showed it to the people at Go SF... It was just kind of a sad open source project, r... |
**Erik St. Martin:** Yeah, it's a beautiful abstraction over the top of doing LXC. |
**Travis Reeder:** Yeah, and it allowed our customers to run and test their code on the exact same environment that it would be running after they uploaded it. That was our initial thing into Docker; we said, "Okay, well now you have Docker, you can test your code before, to fully test it. You could test it locally and... |
And then all of a sudden, okay, now they can run in the exact same environment that they're gonna be running on after it's uploaded. But you'd still upload your code. Then more recently we've just said we'll just run any image, so you can create your own image from the ground up, whereas before we had a bunch of differ... |
**Brian Ketelsen:** So how do you support the any-image plan? Are you piping in and out of standard-in/standard-out? What's the stick behind that one? |
**Travis Reeder:** Yeah, I mean... Even before, we'd just execute your code inside a container, and yeah we'd pull out the logs, store your logs, store the exit status and stuff, but inside, your code does whatever it wants. It can connect to databases and APIs, do some processing and then push the data back into your ... |
**Brian Ketelsen:** That's nice. |
**Erik St. Martin:** You said you didn't have any real battle wounds from Go adoption early, how about containers? |
**Travis Reeder:** \[laughs\] That's a different story, we had a lot of problems with Docker. When Docker first came out there were a lot of bugs. The Docker daemon would lock up, for instance, and basically... In fact it didn't just lock up, it would lock up the entire machine; you couldn't even SSH into it. So there ... |
**Erik St. Martin:** In the late '90s that was my approaching to administering SQL Server - reboot it. |
**Travis Reeder:** Well, it works, you know? It clears out all the crap that's been building up. But it also had memory leaks and things like that, but that also fixed the problem because as long as you restarted the server before the memory ran out... So there were some issues; it's a lot better nowadays, though. |
**Brian Ketelsen:** It really is. Running Docker in production is one of the nicest things about 2016, I can tell you that. Especially with tools like Kubernetes and Rancher and Deis... All of these really nice orchestration tools that make it so easy to manage Docker. |
**Travis Reeder:** I think Docker and these containers are gonna take over the computing world, for sure. I'm pretty bullish on it. |
**Brian Ketelsen:** I agree, too. |
**Erik St. Martin:** So what's your prediction there? You like to make a lot of predictions for Go. What's your prediction for Docker and containerization? |
**Travis Reeder:** Like where it's gonna go? |
**Erik St. Martin:** Yeah. Do you think we'll ever truly get rid of VMs and it will be all containers, all the time? Or they'll just have the primary market share. |
**Travis Reeder:** I think it should be all containers, all the time. I love Rancher, and CoreOS's philosophies - just everything in a container. The beauty of it is you fire up a new server, and all you need to do is have Docker on it, right? |
**Erik St. Martin:** Yeah. I mean, I love CoreOS's approach to the kind of like, you should write crashable software, right? You shouldn't have to plan the server restart, right? At first, that seemed really odd to me, you know? Like, "Oh man, at any point it could update and restart?" That seems crazy. I guess there a... |
**Travis Reeder:** Yeah. We even use it in development a lot now, too. You clone a repo, and without any tooling or anything like that, you can sort of start using it right away, and get into it. |
**Brian Ketelsen:** Yeah, it has made the development workflow... You know, we talked about CockroachDB earlier; CockroachDB had - I don't know if they still do, but last year when we were playing with our fork of it, it had some pretty serious dependencies that had to be installed in order to build it. Making that who... |
**Travis Reeder:** Pretty amazing. On the op side too, it's pretty amazing. We've kind of adopted Docker for all of our deployment and what not, not just for running people's code. Everything's so much easier if you base everything around that one thing. You don't have to say "Okay, well these servers need this stuff i... |
**Erik St. Martin:** Alright, it looks like we're about out of time. I think we actually went over a little bit, but before we go, we have a history of doing kind of like a \#FreeSoftwareFriday shoutout, where all of us will kind of just give thanks and praise to some open source project that is currently or has in the... |
**Brian Ketelsen:** Sure. Last week I've mentioned that I've decided my new year's resolution this year - it's a couple months late. My new year's resolution was to start learning frontend development; I've been doing backend forever and completely left behind by the frontend scene. So I started learning GopherJS, whic... |
**Erik St. Martin:** Carlisia? |
**Carlisia Thompson:** I like to every once in a while acquire new tools to look at my code. I like to know what's going on under the hood, things that I wouldn't necessarily catch just looking at it, and I found this tool called gocyclo. It measures the cyclomatic complexity of your code. There's a simple command line... |
**Erik St. Martin:** Yeah, that's actually really interesting. I have not seen anything like that in Go yet, and cyclomatic complexity measurement is actually really common in the Java world. Almost all CI environments I've ever worked in in Java used it, so that's really cool, I wanna check that out. I don't know whet... |
**Brian Ketelsen:** We have it built into the CI routine in goa, so we will fail if our cyclomatic complexity is higher than 20 on any function. |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.