Kubernetes Pod Disruption Budget (Examples)

[@AntonPutra]

🔴 - To support my channel, I’d like to offer Mentorship/On-the-Job Support/Consulting - [e-mail]

[@AntonPutra]

👉 How to Manage Secrets in Terraform - kzbin.info/www/bejne/aX-TpXqBrNt1mqM 👉 Terraform Tips & Tricks - kzbin.info/www/bejne/bYScZaKLid5lsJY 👉 ArgoCD Tutorial - kzbin.info/www/bejne/sHjRlZqafMZkisU

[@AntonPutra]

🔴Kubernetes Explained Playlist - kzbin.info/aero/PLiMWaCMwGJXkKGx03qyCAqSit6vY5G_JK

[@mmu99]

This short, simple and easy to understand explanation that I need. Thanks Anton

[@AntonPutra]

My pleasure

[@vi5hnupradeep]

Thank you so much . Short and to the point .

[@AntonPutra]

Thanks!

[@pwitty2001]

This was 10x better than the official docs, thank you

[@AntonPutra]

Thanks! Appreciate it!

[@jordotech_tube]

yes, straight to the point, high density and very useful, going to apply this ASAP! Thanks!!!

[@AntonPutra]

my pleasure!

[@ahmedsaif4541]

Hi Anton , thank you very much for you effort , i got the purpose of PDB usage

[@AntonPutra]

Welcome, it is most useful during migrations and cluster upgrades.

[@fartzy]

So what happens if you don’t set PDB and you have three replicas? Doesn’t kubernetes try to keep 3 anyway, what does PDB get you

[@AntonPutra]

Pod Disruption Budgets (PDBs) are useful during Kubernetes node group updates. The 'kubectl drain' command respects your PDB while removing a node from the cluster. For instance, if you have 3 replicas on 1 node, without a PDB, the 'drain' command can evict all of them simultaneously, causing your app to go down. However, with a PDB, 'drain' ensures that a certain number of replicas, as defined in the PDB, continue to run before proceeding. I hope this is helpful.

[@fartzy]

@@AntonPutra yes that is helpful. Is there any other way in which a PDB will be useful other than a node group update where a ‘kubectl drain’ is used? What about an application that is a cronjob with indexed job completion and the job does sharded processing. I have taken over a job like this and I’m seeing a PDB in the helm template (which obviously launches a PDB in k8s). From what you are saying, there is no real purpose to have this PDB there - as I am sure they are not protecting against upgrades and subsequent ‘kubectl drain’ commands

[@AntonPutra]

@@fartzy I don't think it's even applicable for Jobs and CronJobs based on the official doc. However, you can test it locally. Create a minikube cluster with a couple of nodes and try to drain one of the nodes. kubernetes.io/docs/tasks/run-application/configure-pdb/#identify-an-application-to-protect

[@smokzpolski]

Anton, could you please do educational video about setting up and monitoring Istio in K8s?

[@AntonPutra]

Planning to switch to istio sometime in the future :) K8s is a big topic on its own :)

[@smokzpolski]

A bit hard to understand, when we need this feature.

[@AntonPutra]

I'll give you 3 example. First is core-dns that already runs in most clusters. If you terminate all your instances at the same time, no DNS :) Send, zookeeper or etcd need a quorum to properly operate, if you have 3 zks you set pdb to 2. And nginx ingress of cource :)

[@gabecerts4286]

Awesome! as always thanks

[@chasim1982]

Great tutorials, thanks Q. should we use PDB with aws managed nodes?

[@AntonPutra]

of course, if you want to limit disraption

[@diegonayalazo]

Thanks

[@AntonPutra]

my pelasure!