https://rancher.com/ logo
Join Slack
Powered by
This message was deleted.
# neuvector-security
a
This message was deleted.
h
kubernetes distribution and version? what are the resources on the cluster? For neuvector: how many controller and scanner replicas?
c
Distribution: GKE Version: Client Version: v1.30.3 Server Version: v1.29.6-gke.1326000 I do not know what you mean by resources, and there are 3 controllers and 105 enforcers in neuvector
q
I think @hundreds-evening-84071 was asking about available resources in the cluster itself. It might help to know how many rules neuvector is trying. to handle, images in the registry, etc. Are the Controllers running out of RAM? Something else? Where is the “no healthy upstream” message coming from?
c
The thing is that most of the time i can not even access the dashboard, trying to log in results in getting back to the logging screen with either a no healthy upstream message, a session has expired message or upstream connect error or disconnect/reset before headers. reset reason: connection termination message
p
You need to be more specific, are the workload healthy? Also neuvector is a ressource hog in my experience, do you have enough ressources ?
c
It should have enough, it is production environment
p
Yeah i've seen production environments with 150% ram allocation lmao thats not saying anything
c
How can I check it? I am not the one managing the cloud environment, just the one in charge of security
p
Well check if the workloads are healthy
if an OOM reaper keeps sowing, you won't get much productivity out of your pods lmao
c
They are all running without restarts, none have an error
p
All marked healthy?
c
Yes, they are all healthy
i
Just my 2 cent: I would temporarily give the pod with the UI more resources (requested CPU and MEM), i.e. the manager pod. Then, at least, you should be able to see the UI. It could of course also be a problem with your networking. Try to access the UI using port-forwarding.
➕ 1
c
The problem seems to be in our load balancer, as if we access it via port forward it goes smoothly, but when trying to access through the web, we get 503 errors and it is unusable
i
I would open a ticket with google. They have quite good support people and the support doesn't cost much
Likely, you hit some limit which then needs to be adapted to your load/usage
Open in Slack
PreviousNext
Powered by