-
Feature
-
Resolution: Done
-
Critical
-
None
-
Strategic Product Work
-
False
-
-
False
-
OCPSTRAT-848Consistent Ingress/Egress into OpenShift clusters across providers
-
0% To Do, 0% In Progress, 100% Done
-
0
1. Proposed title of this feature request
Support new Azure LoadBalancer 100min idle TCP timeout
2. What is the nature and description of the request ?
When provisioning a service of type LoadBalancer for OCP cluster on Azure, it is possible to customize TCP idle timeouts in minutes using the LoadBalancer annotations 'service.beta.kubernetes.io/azure-load-balancer-tcp-idle-timeout'
Currently, min and max values are hardcoded to respectively 4 an 30 in both legacy Azure Cloud Provider implementation and cloud Provider Azure
Recently Azure upgraded its implementation to support a max of 100 min for idle timeout, corresponding documentation should be updated soon Configure TCP reset and idle timeout for Azure Load Balancer. It is now possible to use such idle timeout with more than 30min manually in Azure portal or with Azure cli but not possible from K8s load balancer as max value is still 30min in K8s code.
Error message returned is
`Warning SyncLoadBalancerFailed 2s (x3 over 18s) service-controller Error syncing load balancer: failed to ensure load balancer: idle timeout value must be a whole number representing minutes between 4 and 30`
3. Why does the customer need this? (List the business requirements here)
Customer is migrating workloads from on premise datacenter to Azure. Using idle timeout with more than 30min is critical to migrate some of our customer links to Azure and is preventing the migration until this is supported by Openshift
4. List any affected packages or components.
Azure cloud controler
- clones
-
RFE-4144 Support new Azure LoadBalancer 100min idle TCP timeout
- Accepted
- links to