Twitter | Search | |
Adrien Trouillaud
Founder and CEO
35
Tweets
36
Following
42
Followers
Tweets
Adrien Trouillaud retweeted
Kelsey Hightower Jun 12
Control planes should be scoped to the smallest failure domain you can afford. A single Kubernetes control plane (think multi regional) is one bad configuration or upgrade away from a regional outage. Multiple zonal clusters, and canary rollouts, is one way to mitigate this.
Reply Retweet Like
Adrien Trouillaud May 23
My talk, "The Toolbox", is already on YouTube! Learn about cross-cluster owner references, cross-cluster garbage collection, and service account imports.
Reply Retweet Like
Adrien Trouillaud May 22
Yesterday, *between* the 1st and 2nd part of my demo at , the cluster and context sections of one of my kubeconfig files mysteriously disappeared. So... who hacked my MacBook? 🕵️ Anyway thanks for attending! I can't believe we had to switch rooms for a bigger one!
Reply Retweet Like
Adrien Trouillaud May 21
544 people and counting have registered for my talk, "The Multicluster Toolbox", today at 3:55pm at , including two ! Thank you all and see you very soon!
Reply Retweet Like
Adrien Trouillaud May 15
Replying to @ibuildthecloud
The purpose of controller-runtime and its derivatives kubebuilder and operator-sdk, IMO, is to make controllers more *approachable*, at the expense of understandability, simplicity, and extensibility. I predict a lot of people will eventually go back to sample-controller.
Reply Retweet Like
Adrien Trouillaud May 12
See -service-account and other multi-cluster tools and design patterns _in action_, by attending "The Multicluster Toolbox", my talk at Europe in Barcelona next week on Tuesday at 3:55pm CEST.
Reply Retweet Like
Adrien Trouillaud retweeted
Brian Grant May 8
Kubernetes Borg/Omega history topic 9: Scheduling constraints. I have volumes more to write about configuration, but will move on with history topics for now. Borg's set of constraints grew organically over time. It started with just required memory, before multicore and NPTL
Reply Retweet Like
Adrien Trouillaud May 7
Replying to @bassamtabbara
Each "thing" individually, or the "things" as a whole? They're Objects of a (Group-Version-)Kind, which maps to a (Group-Version-)Resource at the Kubernetes API. Colloquially, I think we can say that a Deployment is an Object of the Deployments Kind or Resource.
Reply Retweet Like
Adrien Trouillaud retweeted
Admiralty May 2
A strategy would have shielded you from today's global outage. Check out how can help.
Reply Retweet Like
Adrien Trouillaud retweeted
Marco Pracucci 🇪🇺 Apr 28
On why runs many "smaller" Kubernetes clusters - by
Reply Retweet Like
Adrien Trouillaud retweeted
learnk8s Apr 25
Multicluster-scheduler is a system of Kubernetes controllers that intelligently schedules workloads across clusters. In the last release, multicluster-Scheduler integrates with Cilium Cluster Mesh and Global Services.
Reply Retweet Like
Adrien Trouillaud Apr 15
Wrote this a while back for my team at PayScale. It's now live on PayScale's new tech blog:
Reply Retweet Like
Adrien Trouillaud retweeted
crossplane Apr 11
The v0.2 release is now live! Check out all the new features and support for . , and object storage in this detailed blog post: Thanks to the entire community! Onward to v0.3, see you in Barcelona 🤓
Reply Retweet Like
Adrien Trouillaud Apr 7
Replying to @ibuildthecloud
Especially in Go. Dependency injection is non-idiomatic. Makes a tool like controller-runtime difficult to extend. Had to rewrite vs. reuse packages for multicluster-controller.
Reply Retweet Like
Adrien Trouillaud retweeted
Matt Klein Apr 7
I normally shy away from writing about SW engineering, but I decided to write about crashing as a useful mechanism to increase SW reliability. Enjoy!
Reply Retweet Like
Adrien Trouillaud retweeted
Admiralty Mar 19
has everything multicluster-scheduler needed for networking, including cross-cluster health-checking. Excellent blog post by explaining Cilium multi-cluster:
Reply Retweet Like
Adrien Trouillaud retweeted
Admiralty Mar 18
Announcing a new release of -scheduler, v0.3.0, including a new integration with 's cluster mesh and global services.
Reply Retweet Like
Adrien Trouillaud Feb 18
My take: 1) use Helm for packaged, versioned apps, that you consume or produce 2) use Jsonnet for apps that follow templates, e.g., your typical microservice 3) use Kustomize for one-of-a-kind apps by
Reply Retweet Like
Adrien Trouillaud retweeted
Admiralty Feb 16
Announcing a new release of -service-account: v0.3.0. If you've automated the installation and bootstrap steps, you may have noticed a race condition. It's now fixed.
Reply Retweet Like
Adrien Trouillaud Jan 28
Replying to @adrienjt
And for a deeper dive, tune in to the regular SIG Multicluster meeting tomorrow at 9:30am PST (5:30pm UTC), where we'll also discuss multicluster-controller and multicluster-service-account.
Reply Retweet Like