-
Notifications
You must be signed in to change notification settings - Fork 715
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
kubeadm: preflight check false positives in CentOS 7 #6
Comments
From @jberkus on November 6, 2016 21:3 The ebtables false positive does not happen with CentOS 7 Atomic on bare metal. Unfortunately, I don't have a regular Centos 7 on bare metal to test on for comparison. |
From @luxas on November 6, 2016 23:19 Why just don't install ebtables in case you don't have it? Basically kubeadm does just |
From @jberkus on November 7, 2016 0:21 Except that apparently it doesn't, because Kubelet apparently works fine with no ebtables. Also, if ebtables is required, why isn't it marked as a dependency of kubeadm? Or at least mentioned in the docs? |
From @luxas on November 7, 2016 1:22 Yes, ebtables is a dependency of the kubelet, see: kubernetes/kubernetes#28717 and kubernetes/kubernetes#26093. You're right that ebtables doesn't seem to be in the rpm dependencies of the kubelet, sorry for that. |
From @wallrj on November 10, 2016 16:14 FYI I'm seeing the pre-flight check error too:
|
From @luxas on November 11, 2016 21:45 @mikedanese Please push the rpms you built from kubernetes/release#197, so this will be fixed |
From @waynebrantley on November 15, 2016 13:43 etables is easy - just install the package.
|
From @dgoodwin on November 15, 2016 13:52 @waynebrantley fixed but not released yet, details above. |
From @waynebrantley on November 15, 2016 13:55 Ok, I just ran it with --skip-preflight-checks |
From @dgoodwin on November 15, 2016 14:1 Probably very unlikely to be related, this preflight check is just a simple bad assumption that /etc/kubernetes doesn't exist or is completely empty, but the packages create /etc/kubernetes/manifests and the check always fails by default. |
From @jberkus on November 15, 2016 20:13 Unlikely to be related. |
From @mariano-dim on November 21, 2016 17:30 you have to delete the content of /etc/kubernetes and try again |
Just tested the latest packages. False positives appear to be fixed. As such, this issue can be closed. However, I can't close it because I don't own it anymore. |
Ok, thanks @jberkus! |
Because of bug [1] we have to disable preflight checks or node never joins. [1] kubernetes/kubeadm#6 Change-Id: I223f247f7e6cf2dccca068a9db920610a9718349 Signed-off-by: Roman Mohr <[email protected]>
From @jberkus on November 6, 2016 4:30
Platform: CentOS 7.2.1511 on AWS EC2 (t2.medium)
kubeadm version: version.Info{Major:"1", Minor:"5+", GitVersion:"v1.5.0-alpha.2.380+85fe0f1aadf91e", GitCommit:"85fe0f1aadf91e134102cf3c01a9eed11a7e257f", GitTreeState:"clean", BuildDate:"2016-11-02T14:58:17Z", GoVersion:"go1.7.1", Compiler:"gc", Platform:"linux/amd64"}
Symptoms:
While following installation instructions at http://kubernetes.io/docs/getting-started-guides/kubeadm/, I am finding that I need to always disable preflight checks due to spurious failures.
With kubeadmin --init:
This error appears on all CentOS machines on AWS I've tried. It doesn't seem to relate to anything at all that I can find; the only google hit I can get on it is a Kubernetes issue explaining how to use --skip-preflight-checks. It does not seem to block kubeadmin setup.
On the nodes I get an additional error:
Indeed, /etc/kubernetes is not empty ... because the RPMs distributed by Kubernetes.io include a sub-directory in that dir, "manifests".
Copied from original issue: kubernetes/kubernetes#36301
The text was updated successfully, but these errors were encountered: