-
Feature Request
-
Resolution: Done
-
Normal
-
openshift-4.12, openshift-4.12.z
-
None
-
False
-
None
-
False
-
Not Selected
-
-
-
-
1. Proposed title of this feature request
Persist healthy IPv4 and IPv6 connections while SPK(CNF/VNF) pods are being scaled up/down.
2. What is the nature and description of the request?
If an SPK pod is added when OVS is trying to hash(OCPBUGS-29951) this will affect which SPK is selected. This means that if a customer has a TCP connection going from one of the APP pods and then a new SPK pod is added to that namespace; then OVN could switch to another SPK mid-connection.
3. Why does the customer need this? (List the business requirements here)
This can effect performance and reliability while performing maintenance activities such as SPK upgrades (rolling upgrades). This has a greater impact than originally expected.
Furthermore, in the near future the WebScale teams will have the ability to manually scale up and down the SPK pods, which will likely exacerbate the incidence of this issue.
4. List any affected packages or components.
UNK
- is related to
-
FDP-628 ECMP symmetric reply support for connections initiated from behind the route
- Closed