Home / user-507497

Matthias Huttar's questions

Martin Hope
Matthias Huttar
Asked: 2022-08-31 12:10:40 +0800 CST
  • 1

the issue started out of nowhere without any apparent change that we did. All of a sudden, deployments to our AKS (Kubernetes Version 1.23.8) clusters would randomly fail with errors like the below. To my undersatnding this means that the images which should be deployed can't be found in our azure container registry - even though they do exist there.

failed to do request: Head "https://XXX.azurecr.io/v2/XXX/manifests/5510": 
dial tcp: lookup acrobherazueuw1bscdev001.azurecr.io on [::1]:53: 
read udp [::1]:40545->[::1]:53: read: connection refused, rpc error: 
code = Unknown desc = failed to pull and unpack image "XXX.azurecr.io/XXX:5510": 
failed to resolve reference "XXX.azurecr.io/XXX:5510": failed to do request: 
Head "https://XXX.azurecr.io/v2/XXX/manifests/5510": dial tcp: 
lookup XXX.azurecr.io on [::1]:53: read udp [::1]:52124->[::1]:53: read: connection refused]

It's certainly not a firewall issue, and I am 100% confident that the image exists. Even funnier, there's a 20% chance that any random image will actually work and deploy. however, it's also possible that when deployed with two replicas one of them will deploy, the other one will generate that error. Deploying an image that has already been successful deployed previousely will not guarantee a successful deployment.

At this point I am really stuck - I am not finding any good help online for this particular problem. Has anyone else faced that issue before?

Best, Matthias

docker
Martin Hope
Matthias Huttar
Asked: 2021-12-01 01:20:05 +0800 CST
  • 0

Our IT landscape is based on the cloud and utilizes IP whitelisting for access to it. This has been working just fine though all summer as our offices IP addresses are whitelisted. Now that our offices have closed again due to covid, I am working on a solution to allow developers to access all cloud resources from home, without whitelisting private people's IP Addresses.

To solve this problem, I want to implement a VPN, that will tunnel all requests to the services which require whitelisting. When developers are using the VPN, they should all come out at the same public IP which I will then whitelist.

I've been able to setup an Azure VPN with a P2S Configuration and the OpenVPN tunnel type. It is in it's own virtual network that has no peerings to any other virtual networks (because I do not want any connectivity inside the azure network) I've also setup the appropriate IP routes.

However, when I connect to the VPN and try to open our development resources, they do not load but instead socket timeout (seems as if all packets are dropped).

I am quite stuck here, as I can't find any useful information to what could be missing to get this setup to work the way I would want it to work.

I would appreciate any hint and suggestions to make this work.

vpn networking azure azure-vpn
Martin Hope
Matthias Huttar
Asked: 2019-01-30 03:41:12 +0800 CST
  • 1

I'm trying to setup Ingress to serve from a static IP that I have created with gcloud. However it seems that Ingress does not take that configured one but picks a random new one.

The name of that IP Address is homefully-ingress-root and it is 35.234.83.106. My ingress config looks like this: 

apiVersion: extensions/v1beta1
kind: Ingress
metadata:
    name: homefully-ingress
    annotations:
        kubernetes.io/ingress.global-static-ip-name: homefully-ingress-root
    labels:
        app: homefully-ingress
spec:
    # ...

The IP it ends up having, however is 

NAME                HOSTS                       ADDRESS         PORTS     AGE
homefully-ingress   adminpanel.homefully.tech   35.244.164.50   80        12m

It seems like the annotation for the homefully-ingress-root ip address is completely ignored. That IP Address also shows up as "unassigned" in the google cloud console.

Can anybody help me understanding the issue here?

Best, Matthias

kubernetes