antoineco / kovhernetes Goto Github PK

Occasionally, coreos-metadata sets the value of COREOS_OPENSTACK_IPV4_LOCAL to the instance's public IP, probably due to a race.

This breaks commands like kubectl logs or kubectl exec (respectively the containerLogs and exec APIs) and exposes both kubelet and etcd publicly.

# /run/metadata/coreos
COREOS_OPENSTACK_INSTANCE_ID=i-00123456                                  
COREOS_OPENSTACK_HOSTNAME=kovhtestnode01                         
COREOS_OPENSTACK_IPV4_LOCAL=203.0.113.1                                                             
COREOS_OPENSTACK_IPV4_PUBLIC=

$ sudo ss -tlpn
...
LISTEN     0      128       203.0.113.1:10250   *:*   users:(("kubelet",pid=819,fd=19))
LISTEN     0      128       127.0.0.1:2379      *:*   users:(("etcd",pid=877,fd=8))
LISTEN     0      128       203.0.113.1:2379    *:*   users:(("etcd",pid=877,fd=7))

Temporary fix: reboot the nodes.

The tutorial states:

5. A configuration file for the kubectl client tool was written locally

which is more of a wish than a description of what does actually happen as the source contradicts:

# TODO: generate local kubeconfig file

Hi,

I keep getting this issue when trying to deploy kubernetes with this script.

ssh (id = ) does not exist

but nowhere in your docs does it state where I can get the ssh key id from..
tried just using the name I gave it in the manager > got error.
tried using the ID i got from logging into horizon > got same error.
I'm lost on where to find my ssh key id for this script.

can you help?

Opening the API server HTTPS URL in the FIrefox web browser shows the following error:

Secure Connection Failed

An error occurred during a connection to x.x.x.x:6443. Certificate contains unknown critical extension. Error code: SEC_ERROR_UNKNOWN_CRITICAL_EXTENSION 

Chrome does not complain.

I noticed in 1.9 docs that Rescheduler is deprecated
https://kubernetes.io/docs/tasks/administer-cluster/guaranteed-scheduling-critical-addon-pods/#rescheduler-guaranteed-scheduling-of-critical-add-ons

There is a use of scheduler.alpha.kubernetes.io/critical-pod in this project. Consider replacing it with Pods priorities

is it possible with some command?
thanks

Ignition fails to pull kubectl from https://storage.googleapis.com with the error

files: createFilesystemsFiles: createFiles: op(12): GET error: Get https://storage.googleapis.com/kubernetes-release/release/v1.12.0/bin/linux/amd64/kubectl: net/http: TLS handshake timeout

After a while, it eventually succeeds:

files: createFilesystemsFiles: createFiles: op(12): GET result: OK

https://github.com/kubernetes/kubernetes/blob/master/CHANGELOG-1.10.md

Container Linux 1576.4.0
Ignition 0.19.0

cf. https://github.com/coreos/ignition/blob/a8a0a033c526ed4997ee4540db2d3b3047df9b5c/internal/resource/url.go#L255-L267

Unfortunately disabling compression is not an option:
Creating instancesUser data too large. User data must be no larger than 65535 bytes once base64 encoded. Your data is 98916 bytes