[HN Gopher] Understanding ReplicaSet vs. StatefulSet vs. DaemonS...
___________________________________________________________________
Understanding ReplicaSet vs. StatefulSet vs. DaemonSet vs.
Deployments
Author : Abhisman
Score : 51 points
Date : 2025-01-05 13:06 UTC (1 days ago)
(HTM) web link (abhisman.notion.site)
(TXT) w3m dump (abhisman.notion.site)
| ofrzeta wrote:
| Doesn't seem to offer more insights than the Kubernetes docs.
| LetMeLogin wrote:
| Totally. Plus it's misguiding readers with different
| functionality and use cases.
|
| For example "Persistent Storage: StatefulSets provide
| persistent storage to their pods through Kubernetes
| PersistentVolumes".
|
| The thing is that STS doesn't do that. That's actually in POD
| definition.
| MuffinFlavored wrote:
| I had no idea... probably shows I haven't hit an advanced
| enough use case.
|
| > why isn't a statefulset just a deployment/pod with PVC
| mounted?
|
| * StatefulSets provide predictable pod names and hostnames
| (pod-0, pod-1)
|
| * StatefulSets handle pods sequentially (0-1-2), ensuring
| proper cluster initialization.
|
| * StatefulSets maintain a consistent pod-to-PVC mapping even
| after pod rescheduling.
|
| * StatefulSets delete pods in reverse order,
| wbl wrote:
| You can look at the structure in TFA to see this. The only
| discussion of the volume mounting is in the pod template,
| not the Stateful Set itself.
| vanillax wrote:
| You can map a PVC back to a pod without statefulset. You
| just need to create the PV and PVC and map it to a
| deployment.
| sitkack wrote:
| "Come on!"
| vanillax wrote:
| Agree. Its more confusion. PVC / PV are native to k8s not
| something part of statefulsets. In fact you _probably_ only
| need to use Statefulsets for databases. Which is probably
| going to be abstracted away with Operators.
| verdverm wrote:
| It's more about roll forward/backward and services with a
| master+replica or leader election. That PVCs are common in
| stateful sets is more a symptom than the cause
| vanillax wrote:
| Yea for sure. Thats why I dropped the term database
| because thats the exact scenario ( master / replica )
| where pod consistency naming,dns, network etc is
| important for StatefulSet.
| null0pointer wrote:
| Author is most likely farming credibility with blog spam.
| kevmo314 wrote:
| > Understanding the nuances and distinctions between these key
| concepts becomes crucial as you navigate the complex -- yet
| rewarding -- waters of container orchestration.
|
| Can't say I've ever felt rewarded for using Kubernetes, literally
| or metaphorically.
| smitelli wrote:
| Early on I heard the k8s controller ecosystem described as "a
| half dozen concurrent `while true` loops all fighting with each
| other." Can't say I disagree.
| pphysch wrote:
| Isn't that distributed systems in a nutshell?
| ithkuil wrote:
| Isn't that what we all are?
| dgfitz wrote:
| I hope I've avoided containers enough to let this phase pass by.
| Formalizing vocabulary around the different states of a container
| makes me want to throw up my hands in a Seinfeld "come one"
| motion.
| Leo_Verto wrote:
| Like it or not, containers are here to stay. They just make way
| too much sense as an atomic unit for delivering applicstions
| and really aren't that complicated.
|
| Kubernetes is a whole nother beast though and while it
| introduces a ton of overhead, it can be useful at a certain
| organisational scale as an API definition for delivering
| services to other teams.
| crabbone wrote:
| This isn't about the state of container. It's about
| interactions between several containers (a group of containers
| is called a pod). Which, probably, doesn't make it better for
| you, but for the sake of correctness...
___________________________________________________________________
(page generated 2025-01-06 23:01 UTC)