Disqus - Latest Comments for IanMLewis

Re: Container Runtimes Part 4: Kubernetes Container Runtimes & CRI

Ian Lewis — Wed, 04 Dec 2019 19:43:32 -0000

I'm glad it was helpful to you!

Re: Container Runtimes Part 4: Kubernetes Container Runtimes & CRI

Ian Lewis — Sat, 23 Nov 2019 03:22:33 -0000

I'm glad it was helpful! Will look out for you at the meetup.

Re: Container Runtimes Part 4: Kubernetes Container Runtimes & CRI

Ian Lewis — Fri, 31 May 2019 03:25:33 -0000

But of course it has! files never seem to stay put...

Thanks for pointing this out. The link should be fixed now.

Re: Container Runtimes Part 4: Kubernetes Container Runtimes & CRI

Ian Lewis — Thu, 09 May 2019 21:41:38 -0000

Maybe it's this issue? https://github.com/kubernet...

Other than that you might be using an older version of crictl?

Re: Container Runtimes Part 4: Kubernetes Container Runtimes & CRI

Ian Lewis — Fri, 19 Apr 2019 19:33:56 -0000

Glad it was helpful to you!

Re: Container Runtimes Part 4: Kubernetes Container Runtimes & CRI

Ian Lewis — Sat, 13 Apr 2019 17:08:05 -0000

Great to hear! I'm glad you got something out of it.

Re: Container Runtimes Part 4: Kubernetes Container Runtimes & CRI

Ian Lewis — Sat, 13 Apr 2019 17:07:44 -0000

Thanks. I'm glad it was helpful for you.

Re: Container Runtimes Part 4: Kubernetes Container Runtimes & CRI

Ian Lewis — Thu, 21 Mar 2019 18:56:06 -0000

Hi, thanks! It would be great if you translated this into Korean! Please just add my name as original author and link back to the original post.

Let me know when you've finished! :)

Re: Container Runtimes Part 4: Kubernetes Container Runtimes & CRI

Ian Lewis — Thu, 07 Mar 2019 18:17:12 -0000

Grad it was helpful!

Re: Container Runtimes Part 4: Kubernetes Container Runtimes & CRI

Ian Lewis — Thu, 21 Feb 2019 21:01:32 -0000

I'm glad they were helpful!

Re: Container Runtimes Part 4: Kubernetes Container Runtimes & CRI

Ian Lewis — Mon, 28 Jan 2019 23:53:43 -0000

For those paying attention there is also rktlet, which I haven't had time to investigate yet.
https://github.com/kubernet...

Re: Container Runtimes Part 4: Kubernetes Container Runtimes & CRI

Ian Lewis — Mon, 28 Jan 2019 16:30:04 -0000

You're welcome!

Re: The Almighty Pause Container

Ian Lewis — Sat, 29 Sep 2018 00:56:24 -0000

Yah. I guess I wasn't trying to go to deep into the weeds since the post is mostly geared towards folks who don't use the syscalls directly and I cared mostly about wait but I see now that it's necessary. Re-written with full explanation about what exactly fork and exec do and how a new, totally different process is started.

Re: Container Runtimes Part 1: An Introduction to Container Runtimes

Ian Lewis — Wed, 22 Aug 2018 06:38:37 -0000

Thanks! Glad it was useful to you

Re: Container Runtimes Part 2: Anatomy of a Low-Level Container Runtime

Ian Lewis — Wed, 18 Jul 2018 01:56:22 -0000

Thanks! I'm glad you got something out of it.
Also, good catch. I fixed the typo.

Re: Blue/Green Deployments on Kubernetes

Ian Lewis — Sat, 30 Jun 2018 08:20:39 -0000

In general there is a small amount of time between when the service is updated and kube-proxy updates iptables on each node to service the new pods. It should be on the order of a second or two. But could be longer if you have a lot of nodes. I don't think there is an SLO that I know of for updates to services. Either way there shouldn't be downtime as I'm pretty sure the rules are updated atomically.

However, if you can service traffic from your old and new versions of the app at the same time then, by all means, use canaries and rolling-updates.

Re: What are Kubernetes Pods Anyway?

Ian Lewis — Fri, 25 May 2018 21:27:52 -0000

Thanks! Glad you found it useful.

Re: What are Kubernetes Pods Anyway?

Ian Lewis — Wed, 24 Jan 2018 00:45:02 -0000

Yah, a sidecar usually has some sort of interface that is uses to communicate with the main pod. Whether that be a file or a network port over localhost. Theoretically you could replace the sidecar with anything that can implement that interface.

If the containers don't need to communicate in any way then that's a good indication they can run in separate pods. Also if the containers are going to scale out, you would do that by adding new pods rather than new containers onto a single pod. Separate pods should be the default so when in doubt they can be separate I think.

Re: What are Kubernetes Pods Anyway?

Ian Lewis — Tue, 16 Jan 2018 18:15:04 -0000

You're very welcome!

Re: The Almighty Pause Container

Ian Lewis — Mon, 01 Jan 2018 23:43:50 -0000

I'm thinking signals like TERM or HUP could be sent from a sidecar container to another container in the same pod to tell it to restart it to reload it's configuration.

Re: The Almighty Pause Container

Ian Lewis — Sun, 19 Nov 2017 20:09:50 -0000

You're right and I specifically addressed that point in the post. I described it as an entry in the process table. Do you think that I addressed the point inadequately?

Re: The Almighty Pause Container

Ian Lewis — Wed, 15 Nov 2017 22:50:10 -0000

My pleasure!

Re: The Almighty Pause Container

Ian Lewis — Wed, 15 Nov 2017 00:08:17 -0000

All fixed. Make sure to put the -p option before the image name in the command so it's not passed as an argument to the pause process.

Re: The Almighty Pause Container

Ian Lewis — Tue, 14 Nov 2017 23:44:28 -0000

Ah, right, good catch. That's likely a copy and paste error. I'll fix it.

Re: The Almighty Pause Container

Ian Lewis — Tue, 14 Nov 2017 21:18:00 -0000

When you use --net you are joining another containers network namespace so AFAIK you need to expose the ports you want to expose on that container.