@@cloudwithraj Hey Raj would you be willing to do a consultation? For payment of course. My company is stuck on the last piece to launch our server hosting business. Were having issues using the kubernetes JavaScript client to stream pod logs to our react front end. Its been websocket hell. Any help or advice would be awesome and again we can pay.
it was very well explained , i have gone through many videos and docs . but getting more and more confused . specially explanation for targetCPUUtilizationPercentage was so much required . having it in demo is another level .
In the CPU resource requests and limits 0.5 and 1.0 are NOT percentage numbers, they are ABSOLUTE numbers, and express cores. 0.5 means half a core. You need to correct this.
I request you Raj to pls guide me step by step creation on jenkins pipeline ....demo to onboard docker service using jenkins pipeline. How we can mention bitbucket repos inside jenkins pipe and build the same.
Great demonstration. Question, how does the pod updates rolls out in HPA? For instance, I have HPA defined with minReplica's as 2 and max as 10, pods got scaled correctly when deployment triggered. However when I roll out update, it is supposed to bounce both the pods with new config, however I am seeing only 1 pod get bounced and I need to bounce the other one manually. How do I fix this?
Executing kubectl get hpa, is returned 45%/50% with 3 replicas, for instance. This 45%, is the sum of all pods or among these 3 pods, is the pod with highest percentage? Nice explanation, I just would like to understand better this metric.
What if I want to have more than 512 mib for a pod? Do we then need to increase the millicores? Also can we take any amount of millicores like 567 or 666?
on resources section, we are providing request and limit for cpu. A pod will be created with request cpu then how come it grows to use limit cpu ? Kindly clarify how request grows to complete the limit cpu
I have set up my test program very similar to yours. I used apache2 ab command line to increase the load. I was able to see the number of pods increases as the cpu usages goes high. However, there was no time difference in Time per request or Total time taken when I manually increased number of replicas(let's say from replicas=2 to 4) . I played around number of request and number of concurrent requests options from ab command line. I was able to see an improvement in time per request with very particular set of request and concurrent request. I checked CPU usage with htop and it used 100% of CPU in very particular situation. So I had to make conclusion that increasing replicas doesn't always guarantee better performance for user's experience. I want to hear your opinion on this. Thank you
crisp and clear. Saved a lot of my time. Thank you Raj. Simply Brilliant
Thanks!
Wow, TY Ali for the donation, much appreciated.
Beautifully explained. All my queries are cleared. Thank you.
Glad it was helpful Dayananda
Dude, you're a real life super hero. keep it up !
Haha thanks mediatwinkle, glad the video was helpful.
Excellent explanation and very apt presentation with the diagrams. Great work Sir...!
Thanks Thrihesh for the kind words. I am glad you found it useful, thanks for watching.
Excellent Raj ....now my all doubts related to HPA is cleared. Thank You so much
You are most welcome
Raj went beast mode with this tutorial. Very nice work!
Appreciate it!
@@cloudwithraj Hey Raj would you be willing to do a consultation? For payment of course. My company is stuck on the last piece to launch our server hosting business. Were having issues using the kubernetes JavaScript client to stream pod logs to our react front end. Its been websocket hell. Any help or advice would be awesome and again we can pay.
@@byteslinger3187 Ty for the offer! Not doing any consultation currently
All your kubernetes video are really great, can you please share more videos like ingress controler, node affinity
Thanks for the kind words. Couple more Kubernetes videos coming soon!
Awesome Video, really cleared my concept of HPA, and it is working with Cluster Autoscaler. Thank you Very much for the video.
it was very well explained , i have gone through many videos and docs . but getting more and more confused . specially explanation for targetCPUUtilizationPercentage was so much required . having it in demo is another level .
Glad it helped
Very well explained the concepts of autoscaling at different levels
Great explanation. I was able to follow. You saved a lot of time. Thanks
Thanks Raj! You made my day! Great demo!
Glad to hear that!
Great video, can we use prometheus instead metrics server ??
Very clear and well illustrated !.
In the CPU resource requests and limits 0.5 and 1.0 are NOT percentage numbers, they are ABSOLUTE numbers, and express cores. 0.5 means half a core. You need to correct this.
Hi thank you for this video, I just need to ask on 9:47 is it also possible to monitor the targetMemoryUtilizationPercentage in HPA?
Yeah, it's possible. HPA can monitor lot of advanced metrics beyond the basic ones.
spec:
maxReplicas: 8
minReplicas: 2
scaleTargetRef:
apiVersion: apps/v1
kind: Deployment
name: nginx
targetCPUUtilizationPercentage: 60
The explanation is really good, thank you
Glad it was helpful!
I request you Raj to pls guide me step by step creation on jenkins pipeline ....demo to onboard docker service using jenkins pipeline. How we can mention bitbucket repos inside jenkins pipe and build the same.
Great video, really helped me get my head around the HPA concept thanks!
Very welcome!
i guess I am kind of off topic but do anyone know of a good website to watch newly released tv shows online?
@Hassan Harrison Flixportal :)
@Raylan Amir Thanks, I went there and it seems to work :D Appreciate it !
@Hassan Harrison happy to help xD
As always awesome video Raj , Thank you so much ♥
Glad you liked it
absolutely great video sir
Thanks bro, glad you liked it
Great video bro. Thank you very much.
nicely explained.. Thank you so Much!!!
Great presentation!
Thanks so much.
Glad you liked it!
Thanks for detailed explanation!
Glad it was helpful!
u r amazing raj
Excellent explaining
Glad it was helpful!
nice one.. simple and neat
Very good content.
Great demonstration. Question, how does the pod updates rolls out in HPA? For instance, I have HPA defined with minReplica's as 2 and max as 10, pods got scaled correctly when deployment triggered. However when I roll out update, it is supposed to bounce both the pods with new config, however I am seeing only 1 pod get bounced and I need to bounce the other one manually. How do I fix this?
Executing kubectl get hpa, is returned 45%/50% with 3 replicas, for instance. This 45%, is the sum of all pods or among these 3 pods, is the pod with highest percentage? Nice explanation, I just would like to understand better this metric.
Great presentation!
Thanks for watching
Nice Explanation 👍😊
Thanks 🙂
What if I want to have more than 512 mib for a pod? Do we then need to increase the millicores? Also can we take any amount of millicores like 567 or 666?
Loved it.
on resources section, we are providing request and limit for cpu. A pod will be created with request cpu then how come it grows to use limit cpu ? Kindly clarify how request grows to complete the limit cpu
Small correction: mebibyte uses 1024 as base, while megabyte uses 1000 as base.
But what is the role of cpu limit in hpa scalling up and down, do we really need it or we should remove it
where is the reference for the app, like selector shud be there in the HPA config, explanation is indeed good, but i think i missed it
Eks namespace access level via aws sso video also we are looking...
life saver
Can deployment has multiple replica s ?
I mean deployment must have more than one replica to ensure ...!
Wowwwww
I have an error : wget: can't connect to remote host:Connection refused. How can i fix it. My service type is ClusterIP
Thank you so much
You're most welcome
veryy good. thanksssss
Welcome 😊
I have set up my test program very similar to yours. I used apache2 ab command line to increase the load. I was able to see the number of pods increases as the cpu usages goes high. However, there was no time difference in Time per request or Total time taken when I manually increased number of replicas(let's say from replicas=2 to 4) . I played around number of request and number of concurrent requests options from ab command line. I was able to see an improvement in time per request with very particular set of request and concurrent request. I checked CPU usage with htop and it used 100% of CPU in very particular situation. So I had to make conclusion that increasing replicas doesn't always guarantee better performance for user's experience. I want to hear your opinion on this. Thank you
because the purpose is to achieve higher availability, not better performance in speed, HA != faster
thank you
Kube will throttle the pod if cpu usage exceeds limit. Not kill it. There are other circumstances where kube will terminate pods.
How to change default values in hpa
Can you scale based on memory?
Yeah!
why this dashboard get the metrics so long?
omg content.. can we have your Udemy course link? would try to enroll it
bit.ly/3Eku9RH , max discount going on right now, expiring Monday
🆒👍