SessionAffinity v1beta1 cloud.google.com not working [124064870]

Can't Repro

Customer Issue

Status Update

No update yet.

Description

am...@gmail.com

created issue #1

Feb 7, 2019 10:09PM

Please provide as much information as possible. At least, this should include a description of your issue and steps to reproduce the problem. If possible please provide a summary of what steps or workarounds you have already tried, and any docs or articles you found (un)helpful.

Problem you have encountered:
ingress load balancer is not sticky, though i followed tutorial and use advanced gke version.
i implemented confiiguration for session affinity as described on:

https://cloud.google.com/kubernetes-engine/docs/how-to/configure-backend-service (using gke version 1.11.6-gke.6).
i was able to see that the configuration stated that all set fine. as presented below (sessionAffinity: GENERATED_COOKIE, trying IP affinity resulted same):
sudo gcloud compute backend-services describe k8s-be-30123--5094bfdd6e22c063 --global
affinityCookieTtlSec: 50
backends:
- balancingMode: RATE
capacityScaler: 1.0
description: ''
group:

https://www.googleapis.com/compute/v1/projects/card-guard-1299/zones/europe-west1-b/instanceGroups/k8s-ig--5094bfdd6e22c063
maxRatePerInstance: 1.0
connectionDraining:
drainingTimeoutSec: 60
creationTimestamp: '2019-02-07T13:12:51.378-08:00'
description: '{"

kubernetes.io/service-name":"default/cg-nginx1","

kubernetes.io/service-port":"8555"}'
enableCDN: false
fingerprint: yaePVwsmVxA=
healthChecks:
-

https://www.googleapis.com/compute/v1/projects/card-guard-1299/global/healthChecks/k8s-be-30123--5094bfdd6e22c063
id: '2203429310078984636'
kind: compute#backendService
loadBalancingScheme: EXTERNAL
name: k8s-be-30123--5094bfdd6e22c063
port: 30123
portName: port30123
protocol: HTTP
selfLink:

https://www.googleapis.com/compute/v1/projects/card-guard-1299/global/backendServices/k8s-be-30123--5094bfdd6e22c063
sessionAffinity: GENERATED_COOKIE
timeoutSec: 40

What you expected to happen:
i expected that http session will be sticky and will be routed to same backend node, however, it was keep on using round robin fashion

Steps to reproduce:
create kubernetes cluster (version 1.11.6-gke.6), follow the instruction and notice session is not sticky

Other information (workarounds you have tried, documentation consulted, etc):

https://cloud.google.com/kubernetes-engine/docs/how-to/configure-backend-service

Comments

af...@google.com <af...@google.com> #2Feb 12, 2019 06:41PM

Assigned to gc...@google.com.

Hello,

Thank you for reporting this issue.

The GKE team is already aware of this issue and are currently investigating a resolution. At this time, I'm unable to provide you with ETA.
With that being said, any future updates on the progress of the investigation will be posted on this thread. Hence I'll recommend that you star this issue, to remain informed when the issue is resolved.

xa...@shipfix.com <xa...@shipfix.com> #3Apr 24, 2019 12:37PM

Hi,

Any information about the sticky session problem ?

Thanks for your help.

ng...@gmail.com <ng...@gmail.com> #4Jan 6, 2020 06:53AM

Any ETA on this issue?
I think this should have been treated as a basic requirement as other loadbalancers provide.
If the whole design of a product depends on this, then it would be really difficult to use GCE

Message last modified on Jan 6, 2020 06:54AM

ra...@google.com <ra...@google.com> #5May 14, 2020 02:51PM

Accepted by ra...@google.com.

Hello,

Thank you for reaching out. Could kindly confirm if you are facing the same issue in newer supported version (>= 1.14.10-gke.36) of GKE cluster?

ra...@google.com <ra...@google.com> #6May 21, 2020 02:02PM

Status: Won't Fix (Not Reproducible)

Hello,

Thank you for reaching out. Since we didn't hear back from you about the issue, I'll mark this case as "Not reproducible". Please feel free to reopen/create a new one to move forward.

Issue 124064870