Kubernetes Office Hours, 20 Feb 2018

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Kubernetes Office Hours (EU Edition) 20180220

Description

Third Wednesday of every month, everyone is welcome!

http://git.k8s.io/community/events/office-hours.md

A

All right, everybody, how are you my name is Paris Pittman I work at Google, I'm, actually filling in for George today, as he takes a plane back home from San Francisco, thanks again for joining us for today's edition of kubernetes office hours. Today, we're going to answer your user questions live on air with our esteemed panel of experts. You can find us in the office hours channel on slack and check the topic URL for any of the details that you need today. What let's go ahead and introduce ourselves again. My name is Paris.

A

We also have Ilya on the line Eliot. Would you like to go ahead and introduce yourself.

A

Awesome thanks again Elia for always doing this. For us, this is a awesome value, add to the project so before we start really some really quick ground rules. First, one. This is a judgment-free zone, as always, everyone has to start from somewhere. So please help your body out by having a supportive environment in the office, hours, Channel and all channels in the kubernetes project.

A

While we're going to do the best to answer your questions, the panel doesn't have access to your cluster, so live debugging is off topic, but we're going to do your back. Do our best to get you moving down and forward.

A

Ilya you're also encouraged to expand on answers with your experiences and pro tips audience you can help by pasting and URLs to official Doc's blogs or anything that may be relevant to the topic on hand. You can also help us out by tweeting spreading the word and paying it forward. Each session is recorded and available you tube. If there is, if you're, using this as a work resource. Please let us know how we're doing so.

A

We can make it better for you if you want to sit on the panel and spread your knowledge about the kubernetes and the project you're more than welcome, or you can help us take notes as well. I do need a note-taker today. So if there's anybody out there taking notes in the office, our slack channel there is the link to notes feel free to help us out with that.

A

We're also looking for some more marketing help and word-of-mouth so feel free to help us out with that to share on Twitter share on any social media that you're on and, lastly, feel free to hang out in the office our slack channel for help. If you cannot get answers in kubernetes users and various other channels, you're more than welcome to hang out with us there all right. So our first question is up and we have several in the office our slack Channel today.

A

The first question Ilia that I'm seen looks like has to do with API versions, and the person has noticed that deployments have gone from API version: extensions, flash beta, 1 to API version, apps /v 1. Where can we keep track of resources, API versions that need to be upgraded and what should they be upgraded to.

B

It is much less likely all such change to a fewer anytime soon.

A

Fair, alright, and to whoever asked that question feel free to chime in with any add-ons all right wouldn't go ahead to the next question. What are some challenges and best practices for adopting RBC and legacy environments.

B

A

And Ilya there they are in the chat so later feel free to ping them with some more details there all right.

A

Next. What is the acceptance guidance for determining whether to use a single node pole with nodes matching on most extensive workload or multiple new poles with nodes matching the specs of the desired workload, single pull with beefy machines or multiple poles with work load? Specific tuning, never thought I would say beefy machines on own live here. We are.

B

B

But you probably want to talk to somebody.

A

I'm, sorry, oh yeah! What team was that.

B

A

And if you were, if the question holders listening and I'll, also post in the chat after we stop live-streaming, so I'm not banging on the keyboard, but there is AG ke Cooper Nettie slack channel. If you'd like to ask questions in there all right. So next next question now we have for today should I use, kubernetes and all environments or Justin.

B

C

That's actually something.

A

So I, should this user be using kubernetes in continuous integration pipelines examples they want to verify that deploying my application to a cluster works is using mini cube and a CI system to test the sensible.

B

But essentially, there's something that that pulse that have simple, but one other approach could be to use.

B

Going to new main phases from there, so there are different ways to really depends on.

A

We're right and I think the individual said that they cannot join office hours livestream today, so that they might have follow-up questions for you next. How do I share the configuration for accessing my kubernetes cluster example of new developers join the team and they need access to a cluster.

B

A

Thanks Illya and to those that are watching, I am recording, so I can't I have limited typing capabilities. So if there are any other questions, please feel free to drop those in the office hours slack channel on the kubernetes slack instance, alright, and then our next question.

A

Let's see how does kubernetes map volumes specifically volume claims? More specifically static volumes looks like for PVC, there is claim ref, and then you had asked a follow-up question, which is: is it just a local volume like host path and the question holder has NFS and in this case, but they would expect that there might be a common way to map these.

B

It really is based on fire.

B

B

C

Manifesting my sources.

D

B

Thought that you fired this question to have somebody in the next.

A

Alright and alia, I'm actually fishing for our next question here, let's see.

A

Well, first question is: how are you today.

A

Alright, no one I think I. Think I just got one here.

A

B

A

Tears, it's still dark here on the west coast, so I need some eggs here in a second. Let's see: well, I have a question. So if there is one change coming up that you know about in 1.10 that users should be aware of- or maybe something that you think is cool in 110- that users should be aware of what is it.

A

B

B

But that is, that is to simply I have.

C

All this waiting for I'm.

B

Those inside sure of my printer that will be have leaked litter on. He may be looking forward to see me in operational service I'd. So right now, if you want to use a weakness, fine, let's take fine fine! You can't become the same.

A

All right, and then, let's see here um it just saw another one uh I think you actually answered this when we were talking about credentials before, but just in case we didn't and if we did by the way, just go ahead and say that and I'll just tell them to check the recording. But how do you share the kubernetes cluster without passing me? Admin credentials.

B

So this is an approach that we take. I know that do some other things, but we find find it quite insecure to credentials or the content API and, let's see I like using those prevents wrong right and what do.

D

You do if you like.

B

You have to go and that it becomes quite like two years and yeah so I would say that at present it's not particularly easy to issue new users credentials for your Buster on.

B

B

I'm not aware of I know that.

A

All right and then next questions coming to us from our kubernetes users channel the person doesn't have a good DNS set on pods I. They get the nodes DNS setting, but just on my new node, not on my cluster, any idea tried to add a node to their cluster and it seems like they can't use cube DNS stuck on this issue anyhow or exhausts or pointers.

A

This might be a little specific, not necessarily that general.

E

A

Yeah I was about to do the same thing, but if you could, that would be awesome just so I'm not typing in everyone's ear.

A

Okay, in the meantime, though, I'll go ahead and ask you another question: it's about get-ups, actually, oh, and they are, they just joined the office, our slack channel to you by the way, the question from the last one so Joel Ryan in the office hours Locke said. Are there any good examples and links for get-ups, possibly using home? And obviously don't we don't talk about links out loud, but any good examples that you can share verbally with the crew watching.

B

Release it so he.

B

The reason we did this is that, if you use how I'm just out of the box, you end up installing things so I don't actually try out what is it installed in your cluster outside of because tonight's big people cluster, but I'm gonna bring the new cluster for whatever.

B

You it's hard to tell what would only home charts are installed in a trusted advisor? You can cure your.

B

Because it doesn't give you a really will be you into: what's everything installed, it's much easier to actually check it, all on together and I tracking to get before right. We created a third-party resource for custom resource. The money should absolutely be old technologies. That's the resource definition that you can check into it and in down version of the chart.

B

D

We were able to make updates.

B

So the idea really keen to share this with you some tension once it.

A

Sounds exciting everyone's looking forward to it, so it looks like the answer to your question. Previously was the individual using cube, ATM or something else, and they said no, that they have an ansible role: enterprise creation to install the node with different component. Everything seems to be okay on deployment. The node is set ready and pods are well scheduled on it. Let me go back to the original question as well.

A

Here looks like that all the good Deanna set on pods get the nodes DNS setting just not on the new node, just on the new node and not in the cluster. Any idea try to add a node to my cluster and it seems like I, can't use the cube, DNS and they're using ansible any idea on this one. It's really specific. So if not, we can.

B

Yeah I think I may be able to help later on.

A

Okay, yeah: that's why I figured no worries there. Alright, we are still pulling in questions and it looks like I can't believe I'm saying that they use our channel looks somewhat clean.

A

Usually there's always an unanswered question in here. Let's see.

A

Okay, um I: don't think this individual got an answer. If they did, then oh Allah will answer alive again on air. Doesn't matter what you name, cease: monitoring, metrics logging apps are deployed, or is it okay to leave in default namespace, or should they all be in the namespace together? What are the pros? The cons.

B

A

It's in the users channel and if you scroll up, if you scroll up a few above your posting today about it, you'll see it, they did not go into any more details, so they're just pretty much asking about default namespaces and if they should be all on the same nice namespace together and the pros and the cons.

B

It's it's really recommended to make good use of namespaces. There are no I mean first things places like.

D

B

What we do it B works, we use namespaces and we use namespaces for four for each of the subsets of.

B

Time series they face for other pollutant service.

B

For example, we.

D

Have the monitoring.

B

Component, like an extra means nice extra from the bits that are most critical, and we also have few things.

B

So this is kind of degree that we organize our services in bridges. Basically, yes, we have a name space, but just seems to be some functionality and each of them namespaces is usually mostly independent from from one another. So let's say everything running and they're, monitoring, namespace Luke said like like Prometheus, etc, is something that oh.

A

Thanks, oh yeah, and this question is surrounding the: how of control loops, so the actually picked out a quote from from a piece of our Docs and the quote: is a control loop in the master, watches for new PVCs finds a matching TV if possible and binds them together, and the question asker would like to know how like how is that? How is that going on? What's going on there.

B

So this is very specific to PVCs I think.

B

Well, it's like.

B

Diseases then give me smiling finesse storage. You don't. Sicily have to have a.

B

You start or or.

B

You have so essentially PVCs. That is something that we talk to your provider, so we provided creates.

B

As you specified.

B

A

Would be what would be a good contact, or maybe a better slot channel for that question?.

A

All right, and then we have another, how kind of question- and this is how volume claims are actually working.

A

They said that they found some information in our Doc's about persistent volume, but no clue how to actually use them. This is probably a storage question, but you have any insight here.

B

B

Depravation geeky specific recommendation in the GCP box, so yeah I was struggling to pass the advance, so I I can hear.

D

Your pain and I would.

B

D

B

To six story.

A

Alright, it looks like they went on to explain some more in the chat and that they know how to use volume, claims and templates, but not how to mount such a static, persistent volume and from what they get that they have to use a volume clean, because that's the only thing that can mount in a pod. So they need a volume clean, that's a filled by the volume, and there seems to be not many docks on that piece. So we should just so.

A

If you want I can just let them noted to visit sig to visit sig storage and see what they say about that.

A

Alright, next question: it looks like someone has a pod that exposes REST API and it's secured by basic off. They want to add a readiness probe. Is there any option to use environment as a value of a header in the readiness probe.

A

Looks like they also tried it with exec probe as well and could not receive any access to the pod environment and that the environment variables not get said tried to use interpolation in executive script using pod environments.

A

This also seems pretty specific as well, and I am also trying to get our next question.

A

Did you say you are trying to answer that? I'm, sorry, okay,.

A

Okay, all right I am looking for our next question here.

A

Ari so far, not much.

A

Anything else exciting in the land of kubernetes that you'd like users to know about. Well, you have a couple minutes and will I fish for our next question.

B

A

I think they did the full schedules out: I believe they're, yeah, yep I think there are some missing key notes, but I believe the full schedule is out.

B

A

I am also extremely excited as well. This will be my first time in Copenhagen and not my first cute con, but very excited to see everyone all right. I just put a last call for questions out there. It looks like kubernetes users might have some more stuff going on. Let's see here.

A

No I don't really see anything at to see a lot of very specific questions and then more stuff related to the storage questions. How we were asking earlier.

A

It's all good, we will get this straightened out and.

A

Alright, so we just always want to say quick things to all of the companies that allow and other contributors to spend their time on office hours so folks in particular, VMware we work red, pivotal pocket, Northwestern, Mutual, liquid Web, fto giant swarm that Namie Amazon Google soon I know George, who was the regular host, is going to be doing some swag giveaways and things like that for question. Askers so feel free to drop back into office hours. As always, our office hours information is posted in the kubernetes lac instance.

A

Hashtag office hours feel free to hang out in the chat or hang out really anywhere with us. Okay, alright, so we are wrapping. This call I am ending this a meeting and we are stopping the stream. Thank you. So much I really enjoyed my first time, quick, plug, really quick plug for meet our contributors, which happens every first Wednesday of the month. This is a user if this is a office hours type session for contributors.

A

So if you are a contributor listening to the call or someone that you someone that would like to push code upstream or docks for that matter, feel free to join our meet our contributors slack channel. Alright, thanks so much.

B

A

Yeah London get your kubernetes on all right, bye. All.