Kubespray: vagrant up not working

Created on 22 Nov 2018 · 15Comments · Source: kubernetes-sigs/kubespray

BUG

I clone the repo, change the OS to centos and then run pip install -r and then vagrant up.

This is the output.

Is there something i am doing wrong?

Thank you,
T

Environment:

Cloud provider or hardware configuration: VAGRANT

OS (printf "$(uname -srm)\n$(cat /etc/os-release)\n"): macOS
Version of Ansible (ansible --version): 2.7.2

Kubespray version (commit) (git rev-parse --short HEAD): 0cfcd39d

Network plugin used:
default

Copy of your inventory file:
??

Command used to invoke ansible:
vagrant up

Output of ansible run:

```==> k8s-3: Running provisioner: ansible...
Vagrant has automatically selected the compatibility mode '2.0'
according to the Ansible version installed (2.7.2).

Alternatively, the compatibility mode can be specified in your Vagrantfile:
https://www.vagrantup.com/docs/provisioning/ansible_common.html#compatibility_mode

k8s-3: Running ansible-playbook...

SUDO password:
/Users/christophe.trefois/dev/test/kubespray/inventory/sample as an inventory
source

that the implicit localhost does not match 'all'

PLAY [localhost] ********************
skipping: no hosts matched
PLAY [bastion[0]] *******************
skipping: no hosts matched
PLAY [k8s-cluster:etcd:calico-rr] **************
skipping: no hosts matched

PLAY [k8s-cluster:etcd:calico-rr] **************
skipping: no hosts matched

PLAY [etcd] **********************
skipping: no hosts matched

PLAY [k8s-cluster:calico-rr] *****************
skipping: no hosts matched

PLAY [k8s-cluster] *******************
skipping: no hosts matched
PLAY [kube-master] *******************
skipping: no hosts matched

PLAY [k8s-cluster] *******************
skipping: no hosts matched

PLAY [kube-master[0]] ******************
skipping: no hosts matched

PLAY [kube-master] *******************
skipping: no hosts matched

PLAY [calico-rr] *********************
skipping: no hosts matched

PLAY [k8s-cluster] *******************
skipping: no hosts matched

PLAY [kube-master] *******************
skipping: no hosts matched

PLAY RECAP ***********************

Thursday 22 November 2018 12:09:45 +0100 (0:00:00.083) 0:00:00.083 *
===============================================================================```

Source

Trefex

Most helpful comment

I can confirm nuking the inventory/sample/hosts.ini gets things moving. Thanks @jhalfmoon

easel on 27 Nov 2018

👍16 🎉1

All 15 comments

cc @jhalfmoon ^

ant31 on 22 Nov 2018

Confirmed. My cleanup of the vagrant file changed a reference to a non-existent file: hosts. The vagrantfile now actually references an existing file, hosts.ini, ironically breaking the vagrant config. The quick fix for this is to rename or remove inventory/sample/hosts.ini before you start the vagrant run. I'll do a PR as soon as I've tested of this. The reason I didn't catch this error earlier is because I use a script to test my changes and in this script I had an 'rm hosts.ini', for test purposes, which I had forgotten about.

jhalfmoon on 22 Nov 2018

I can confirm nuking the inventory/sample/hosts.ini gets things moving. Thanks @jhalfmoon

easel on 27 Nov 2018

👍16 🎉1

This solution worked for me. Is this still open (not merged)? I thought it would have landed in the latest release.

tdensmore on 4 Jan 2019

Even after deleting the inventory/sample/hosts.ini file, you still need to enter the password of the local user. This is not in the docs.

rgevaert on 20 Jan 2019

👍3

@jhalfmoon which PR would that be?
Is this still a problem?

Miouge1 on 7 Feb 2019

This is definitely still a problem @Miouge1. First time using kubespray to setup a local multi-node dev environment and I'm getting this:

Version: master branch an hour ago.

==> k8s-3: Running provisioner: ansible...
Vagrant has automatically selected the compatibility mode '2.0'
according to the Ansible version installed (2.7.7).

Alternatively, the compatibility mode can be specified in your Vagrantfile:
https://www.vagrantup.com/docs/provisioning/ansible_common.html#compatibility_mode

    k8s-3: Running ansible-playbook...
SUDO password:
 [WARNING]: provided hosts list is empty, only localhost is available. Note
that the implicit localhost does not match 'all'


PLAY [localhost] ***************************************************************
skipping: no hosts matched
 [WARNING]: Could not match supplied host pattern, ignoring: bastion


PLAY [bastion[0]] **************************************************************
skipping: no hosts matched
 [WARNING]: Could not match supplied host pattern, ignoring: calico-rr


PLAY [k8s-cluster:etcd:calico-rr] **********************************************
skipping: no hosts matched

PLAY [k8s-cluster:etcd:calico-rr] **********************************************
skipping: no hosts matched

PLAY [k8s-cluster:etcd:calico-rr] **********************************************
skipping: no hosts matched

PLAY [etcd] ********************************************************************
skipping: no hosts matched

PLAY [k8s-cluster:calico-rr] ***************************************************
skipping: no hosts matched

PLAY [k8s-cluster] *************************************************************
skipping: no hosts matched

PLAY [kube-master] *************************************************************
skipping: no hosts matched

PLAY [k8s-cluster] *************************************************************
skipping: no hosts matched

PLAY [kube-master[0]] **********************************************************
skipping: no hosts matched

PLAY [kube-master] *************************************************************
skipping: no hosts matched

PLAY [calico-rr] ***************************************************************
skipping: no hosts matched

PLAY [k8s-cluster] *************************************************************
skipping: no hosts matched

PLAY [kube-master] *************************************************************
skipping: no hosts matched

PLAY RECAP *********************************************************************

dotnwat on 8 Feb 2019

The ini extension is ignored in ansible.cfg so an inventory relying on an .ini file can't be supplied as a directory.

Looks like #4153 resolves the issue for the Vagrantfile.

Why are .ini files ignored for inventory in the first place if that's what the examples use?

mhio on 11 Feb 2019

Looks like #2446 added the inventory_ignore_extensions line with more than just .creds. Maybe ini should be removed?

mhio on 11 Feb 2019

It's the Ansible default, see https://github.com/ansible/ansible/issues/39562#issuecomment-385696800:

This is on purpose, due to the legacy of inventory scripts and their reliance on .ini files for configuration. If you point at the file directly or change the configuration it will work, but if we change the default it will break many existing setups that rely on the .ini files being ignored. Also the ini plugin can parse files w/o the .ini extension.

And https://github.com/kubernetes-sigs/kubespray/pull/2446#discussion_r175597591

manics on 11 Feb 2019

I had same issue when trying vagrant up at the first time. To make this easier, is it possible
to remove or rename hosts.ini file under inventory/sample? Renaming like
hosts_sample.ini might be easy to understand the purpose.
This repo is great for vagrant users, thanks to all.

kobayashi on 1 Mar 2019

Just followed this comment: https://github.com/polyaxon/polyaxon/issues/349#issuecomment-462052351 and vagrant up worked as charm.

Also I had to upgrade to latest Vagrant (from 2.0.2 to 2.2.4), since I got ruby errors along the way, eg. #<Thread:0x0000556fc7ef62a8@/usr/share/rubygems-integration/all/gems/vagrant-2.0.2/lib/vagrant/batch_action.rb:71 run> terminated with exception (report_on_exception is true): Traceback (most recent call last): 107: from /usr/share/rubygems-integration/all/gems/vagrant-2.0.2/lib/vagrant/batch_action.rb:82:in `block (2 levels) in run' 106: from /usr/share/rubygems-integration/all/gems/vagrant-2.0.2/lib/vagrant/machine.rb:188:in `action...

Hope it helps.