You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Could you please enhance CloudStack to enable support for multiple ports in autoscaling? The current limitation of only one port restricts the deployment of real-time production infrastructure with autoscaling groups.
EXPECTED RESULTS
To address this issue of limited ports in the next release!
at first sight this looks like a valid feature request @palashbiswas-git , but can you elaborate on the configurations/requests/settings that you expect and how they would bahave, including error handling.
I'm currently working with a container-based application that supports autoscaling. However, I'm facing a limitation due to having a single load balancer and single port, which prevents me from parking it under an Autoscaling group, especially when I have multiple containers in a single VM/Instance. As you know, except for frontend web-based applications, most applications work with multiple ports where the VM needs to connect to other corresponding nodes, containers, and other services.
Because of this limitation, I believe many Cloudstack users are unable to secure those backend critical systems/applications through autoscaling groups.
Therefore, it would be greatly appreciated if this feature could be included in an upcoming enhancement.
I hope this clarifies the issue and the configuration/setting we are looking for in the future.
You are correct that this can be achieved through Kubernetes. However, our requirement is to use either containers or normal VMs, where we can't handle multiple load balancers or a single load balancer with multiple ports if it's part of an autoscaling group.
This is a great feature for us. Right now, in order to overcome this limitation, we deploy an NGINX instance in our VM, configured with static routes. These static routes would then route traffic to the multiple dockers in the VM.
Then, we scale the entire VM altogether.
Its a pretty hacky workaround, but it helps simplify our deployment.
Autoscaling with multiple ports would be really great! I full support.
@palashbiswas-git , I am sure you have a valid use case, and I think something like this could be implemented in the future. I read this as a generic problem statement and not a very explicit one. I am moving this to discussion to further distill it and create (probably multiple) issues once it is more clear.
Activity
DaanHoogland commentedon Feb 7, 2024
at first sight this looks like a valid feature request @palashbiswas-git , but can you elaborate on the configurations/requests/settings that you expect and how they would bahave, including error handling.
palashbiswas-git commentedon Feb 28, 2024
Hi @DaanHoogland ,
I'm currently working with a container-based application that supports autoscaling. However, I'm facing a limitation due to having a single load balancer and single port, which prevents me from parking it under an Autoscaling group, especially when I have multiple containers in a single VM/Instance. As you know, except for frontend web-based applications, most applications work with multiple ports where the VM needs to connect to other corresponding nodes, containers, and other services.
Because of this limitation, I believe many Cloudstack users are unable to secure those backend critical systems/applications through autoscaling groups.
Therefore, it would be greatly appreciated if this feature could be included in an upcoming enhancement.
I hope this clarifies the issue and the configuration/setting we are looking for in the future.
Thank you.
kiranchavala commentedon Feb 28, 2024
Hi @palashbiswas-git
Have you tried cloudstack autoscaler for kubernetes ?
https://github.com/kubernetes/autoscaler/tree/master/cluster-autoscaler/cloudprovider/cloudstack
When a Kubernetes pod scales up , cloudstack automatically provisions a node
apache/cloudstack-kubernetes-provider#52 (comment)
palashbiswas-git commentedon Feb 28, 2024
Hi @kiranchavala,
You are correct that this can be achieved through Kubernetes. However, our requirement is to use either containers or normal VMs, where we can't handle multiple load balancers or a single load balancer with multiple ports if it's part of an autoscaling group.
btzq commentedon Mar 4, 2024
This is a great feature for us. Right now, in order to overcome this limitation, we deploy an NGINX instance in our VM, configured with static routes. These static routes would then route traffic to the multiple dockers in the VM.
Then, we scale the entire VM altogether.
Its a pretty hacky workaround, but it helps simplify our deployment.
Autoscaling with multiple ports would be really great! I full support.
DaanHoogland commentedon May 31, 2024
@palashbiswas-git , I am sure you have a valid use case, and I think something like this could be implemented in the future. I read this as a generic problem statement and not a very explicit one. I am moving this to discussion to further distill it and create (probably multiple) issues once it is more clear.