Fails to add status to ingress after running "for some time" #3180
Labels
triage/needs-information
Indicates an issue needs more information in order to work on it.
Comments
|
@cjohansen if you see this behavior again, please update the issue running: and post the output of: This can help us to narrow where this issue is being triggered. |
aledbf
added
the
triage/needs-information
|
Thanks! I haven't seen this again, but I still have one of the old (presumed "bad") pods running. Would it be any use to get the output of this now? |
Yes please :) |
|
http://localhost:10254/debug/pprof/ http://localhost:10254/debug/pprof/goroutine?debug=1 http://localhost:10254/debug/pprof/block?debug=1 |
|
I also had to restart the nginx-ingress-controller. Previously it was cycling through reloads and not finding any active Endpoints, as described above. This was on a fresh install on Docker for Mac following the installation steps on your Deploy page (mandatory and cloud-provider yamls). Unfortunately I restarted it before reading your request for debug information. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
NGINX Ingress controller version: 0.18.0
Kubernetes version (use
kubectl version): 1.10.3Environment:
uname -a): 4.4.121-k8sWhat happened
It's about a month since I set up Kubernets with the ingress controller. At the time I created a few services, and everything worked as expected. Sat down to deploy a new service today, and noticed that no load balancer was connected to the ingress. Checked the ingress controller's logs and found:
Checked
kubectl get epand saw:Double and triple checked my template, deleted all the resources and retried. Unfortunately I am not entirely sure if I made any changes that matter at this point. Still no luck. Log now reads:
I eventually decided that this simply was not correct, so I decided to try to kill one of two ingress pods. Shortly after the new one booted up, the ingress was properly configured. The new pod's log now reads:
What you expected to happen:
Not needing to reboot the nginx ingress controller in order for my ingresses to be correctly configured.
How to reproduce it (as minimally and precisely as possible):
Unfortunately I do not know. Possibly: leave the ingress controller running for 20+ days without any new events to respond to, then try to create a new service+ingress.
Anything else we need to know:
I'm sorry for reporting something that is probably very hard to follow up on. I'm not 100% sure this wasn't some sort of mistake on my behalf, but I'm reporting anyway since rebooting one of the ingress-nginx pods fixed the problem, which was somewhat surprising. I'm hoping someone who know the internals better might have a eureka moment from this report...
I still have one of the old pods running, and would happily provide more logs etc if needed.
The text was updated successfully, but these errors were encountered: