Sir, Here how you are increasing the pods like what are the metrics you are using. Or setting a target value to compare with the currentvalue. I mean to make the scaling up and down based on some targetvalue if the request rate is increased than the targetvalue then the scaling up or else scale down.
Hi, i am using the gateway and httproute for my own domain and my app is up and running in prod but i want to use keda so there a way i can do that without the portfoward by which means i simply just hit the domain from the browser and the app will scale up?
@@vaibhavverma2579 yes you can do that port forward is just for demo here in real applications are deployed properly using loadbalancer/ingress in k8s cluster.
@@rohanrustagi7857 i am not using ingress i have creates the wildcard domain in the gateway and setup a httproute which connects to my app's svc and svc to pods so how can i scale from 0 to 2 or vice versa? Can u guide me on that pls because rightnow without portfowarding my app is not scaling up to 1 again.
@@rohanrustagi7857 ia there any way to do without portforwarding because right now its not scaling up to 1 again with prot forward i have my http route and gateways configured so i have my domain as well
Hi Sir,.. thanks for the session. I have some doubts, for testing purpose do we need to purchase a domain for this request based scaling activity ? Can't we use ip with nodePort for testing with request based scaling ? And how to make sure that at what point my scaling process will be started, is it not necessarily to be there to configure a set request and to confirm if it reached more than defined request then pods will be created or else go down just like we set utilisation for cpu/memory. Once pod receive signal to go down then what about existing request, will those request going to be complete before pod move to termination or directly it will go down. And what about upcoming request will those http request will be go to other pod or simply it goes to the same pod that going to be terminated?
1. No need to purchase domain because even in the official github repo of keda they even use a dummy domain name. 2. Yes you can use nodeport for testing 3. In http addon , everything is based on http request , cant say about cpu/mem. Request here we can use such as cooldownperiod , fallbackoption,etc. 4. Yes the external scaler always make sure that every request is fulfilled , even if last request is not completed the pods will be up. No termination will happen till the last. 5. For handling request by pods that is dependable upto you as projects use loadbalancer with ingress.
thankyou very much sir for your support. We will always be grateful for you. We are using kubeadm environment with one Master 2node, If you are planning for a session or anywhere like comment section please update us what would be the format if we use ip with nodeport in host section instead of domain.
@@rohanrustagi7857sir, please create a session for number of http request based autoscalling to deployment service with Prometheus, Prometheus adapter using helm for kubeadm environment. I am eagerly waiting for it..
Sir,
Here how you are increasing the pods like what are the metrics you are using. Or setting a target value to compare with the currentvalue. I mean to make the scaling up and down based on some targetvalue if the request rate is increased than the targetvalue then the scaling up or else scale down.
Yes if currentvale > target value then : scale up
Else scale down. Also to mention max replica and min replica in the keda yaml manifest file.
Hi, i am using the gateway and httproute for my own domain and my app is up and running in prod but i want to use keda so there a way i can do that without the portfoward by which means i simply just hit the domain from the browser and the app will scale up?
@@vaibhavverma2579 yes you can do that port forward is just for demo here in real applications are deployed properly using loadbalancer/ingress in k8s cluster.
@@rohanrustagi7857 i am not using ingress i have creates the wildcard domain in the gateway and setup a httproute which connects to my app's svc and svc to pods so how can i scale from 0 to 2 or vice versa? Can u guide me on that pls because rightnow without portfowarding my app is not scaling up to 1 again.
@@rohanrustagi7857 ia there any way to do without portforwarding because right now its not scaling up to 1 again with prot forward i have my http route and gateways configured so i have my domain as well
Hi Sir,.. thanks for the session. I have some doubts, for testing purpose do we need to purchase a domain for this request based scaling activity ? Can't we use ip with nodePort for testing with request based scaling ?
And how to make sure that at what point my scaling process will be started, is it not necessarily to be there to configure a set request and to confirm if it reached more than defined request then pods will be created or else go down just like we set utilisation for cpu/memory.
Once pod receive signal to go down then what about existing request, will those request going to be complete before pod move to termination or directly it will go down. And what about upcoming request will those http request will be go to other pod or simply it goes to the same pod that going to be terminated?
Please give a solution to this, if possible using KEDA or else if any other process is there to achieve the above using the Prometheus adapter
1. No need to purchase domain because even in the official github repo of keda they even use a dummy domain name.
2. Yes you can use nodeport for testing
3. In http addon , everything is based on http request , cant say about cpu/mem. Request here we can use such as cooldownperiod , fallbackoption,etc.
4. Yes the external scaler always make sure that every request is fulfilled , even if last request is not completed the pods will be up. No termination will happen till the last.
5. For handling request by pods that is dependable upto you as projects use loadbalancer with ingress.
I am trying to make video for prometheus scaler as well in which based autoscaling can start on basis of http_requests
thankyou very much sir for your support. We will always be grateful for you.
We are using kubeadm environment with one Master 2node, If you are planning for a session or anywhere like comment section please update us what would be the format if we use ip with nodeport in host section instead of domain.
@@rohanrustagi7857sir, please create a session for number of http request based autoscalling to deployment service with Prometheus, Prometheus adapter using helm for kubeadm environment. I am eagerly waiting for it..