Skip to content
This repository has been archived by the owner on Apr 29, 2019. It is now read-only.

VM use DHCP instead of static IP adresses as expected #47

Closed
ggarson opened this issue Mar 18, 2015 · 10 comments
Closed

VM use DHCP instead of static IP adresses as expected #47

ggarson opened this issue Mar 18, 2015 · 10 comments

Comments

@ggarson
Copy link

ggarson commented Mar 18, 2015

Hi,

before the commit 7f33709 everything was OK.

Now when I deploy a cluster with virtualbox provider (I configured a DHCP serveur inside virtualbox in the subnet 172.17.8.x-49/99), nodes use IP provided by the DHCP instead of 172.17.8.101 (master) and so on for minions.

P.S. I don't used the sync file, so all is by default.

Tell me if you need more information to reproduce the problem.
@AntonioMeireles
Copy link
Contributor

@ggarson 😢
afaict none of the changes at the above commit in the Vagrantfile should ever change the way the vagrant/hypervisor networking was going on.

so, the issue persists if you just disable (per the README) the 'native' VirtualBox DHCP ? ( VBoxManage dhcpserver remove --netname HostInterfaceNetworking-vboxnet0 )

@ggarson
Copy link
Author

ggarson commented Mar 18, 2015

In fact there is no issue if I disable it (I did it as a workaround) but before it was working event if the DHCP was enabled.

In fact I use another virtual machine (or more) in virtualbox in the subnet 172.18.8.0 to access directly to my coreos host. And there virtual machines are configured to use a dhcp.

So it's not really an issue but more precisely a regression. Someone was faced of the same problem than me on irc chan of Kubernetes.

@pires
Copy link
Owner

pires commented Mar 18, 2015

@ggarson @AntonioMeireles there are many issues coming up with that merge and I'm currently removing them, but since the PR had so many things - my bad! - it's hard to revert. Am working on it as we speak.

@AntonioMeireles
Copy link
Contributor

i believe you i just have no freaking idea of what triggered this change of behaviour. will digg. (btw if you use old CoreOS alpha images - issue persists ? (just to make sure that this isn't some upstream change)

@ggarson
Copy link
Author

ggarson commented Mar 18, 2015

I use the last alpha images for days and I was only a problem this morning after the git pull... I reinstall my clusters few times by day

@pires
Copy link
Owner

pires commented Mar 18, 2015

@ggarson working on a hotfix as we speak. Please bear with me ;-)

@ggarson
Copy link
Author

ggarson commented Mar 18, 2015

@pires don't worry, you are doing a great job ;)

pires added a commit that referenced this issue Mar 18, 2015
@pires
Desperate attempt at fixing so many issues that were brought by what …
559df1a 
…we thought were smart moves. Sorry for that!

Removing dockerized etcd 2.x
Removing dockerized flannel
Master node doesn't need to rely on its own dockercache for pulling images, for now.
Refs #46 #47 #32
@pires
Copy link
Owner

pires commented Mar 18, 2015

@ggarson thanks for your kind words. Can you please try it out now and let me know?

@ggarson
Copy link
Author

ggarson commented Mar 18, 2015

It's OK for the DHCP issue.

Thanks @pires

@ggarson ggarson closed this as completed Mar 18, 2015
@pires
Copy link
Owner

pires commented Mar 18, 2015

Awesome @ggarson! Thanks for your feedback man.
On Mar 18, 2015 19:38, "ggarson" notifications@github.com wrote:

It's OK for the DHCP issue.

Thanks @pires https://github.com/pires

—
Reply to this email directly or view it on GitHub
#47 (comment)
.

zutherb pushed a commit to zutherb/kubernetes-vagrant-coreos-cluster that referenced this issue Mar 28, 2015
Merge branch 'master' of git://github.com/pires/kubernetes-vagrant-co…
d776b6e 
…reos-cluster into pires-master

* 'master' of git://github.com/pires/kubernetes-vagrant-coreos-cluster: (35 commits)
  Fixes pires#51
  Allow for running privileged containers.
  Warning, only Kubernetes 0.13.x or newer is currently supported (and working).
  Removing etcd debugging stuff in order to simplify things.
  Ignoring bad SSL certificates when downloading kubectl.
  Fixing typos.
  Kubernetes upgraded to 0.13.2.
  kube-proxy is deprecating flag --ectd_servers in favor of --master. Refs pires#50
  Reverted to Kubernetes 0.12.2 since 0.13.x has broken services iptables rules. See kubernetes/kubernetes#5647
  Added waiter for Kubernetes API server before starting kube-proxy and kubelet.
  Latest Kubernetes version is really the latest tested, so 0.13.1 now.
  Desperate attempt at fixing so many issues that were brought by what we thought were smart moves. Sorry for that! Removing dockerized etcd 2.x Removing dockerized flannel Master node doesn't need to rely on its own dockercache for pulling images, for now. Refs pires#46 pires#47 pires#32
  Added more hypervisor related documentation love, namely removing DHCP from Virtualbox. Fixes pires#45
  Removed incompatible Docker flags because of bad merge.
  drop unneeded nfs related cruft.
  docker-cache container now restarts on failure. Refs pires#40
  only start docker in the minion nodes after the master's docker-cache is up.
  adding workaround for docker-archive/docker-registry#892
  rollback 182252c. just default kubernetes to latest stable release.
  fix pires#36 - docker could start before flanneld
  ...
zutherb pushed a commit to zutherb/kubernetes-vagrant-coreos-cluster that referenced this issue Mar 28, 2015
Merge branch 'pires-master'
6cb593a 
* pires-master: (35 commits)
  Fixes pires#51
  Allow for running privileged containers.
  Warning, only Kubernetes 0.13.x or newer is currently supported (and working).
  Removing etcd debugging stuff in order to simplify things.
  Ignoring bad SSL certificates when downloading kubectl.
  Fixing typos.
  Kubernetes upgraded to 0.13.2.
  kube-proxy is deprecating flag --ectd_servers in favor of --master. Refs pires#50
  Reverted to Kubernetes 0.12.2 since 0.13.x has broken services iptables rules. See kubernetes/kubernetes#5647
  Added waiter for Kubernetes API server before starting kube-proxy and kubelet.
  Latest Kubernetes version is really the latest tested, so 0.13.1 now.
  Desperate attempt at fixing so many issues that were brought by what we thought were smart moves. Sorry for that! Removing dockerized etcd 2.x Removing dockerized flannel Master node doesn't need to rely on its own dockercache for pulling images, for now. Refs pires#46 pires#47 pires#32
  Added more hypervisor related documentation love, namely removing DHCP from Virtualbox. Fixes pires#45
  Removed incompatible Docker flags because of bad merge.
  drop unneeded nfs related cruft.
  docker-cache container now restarts on failure. Refs pires#40
  only start docker in the minion nodes after the master's docker-cache is up.
  adding workaround for docker-archive/docker-registry#892
  rollback 182252c. just default kubernetes to latest stable release.
  fix pires#36 - docker could start before flanneld
  ...
Hikariii pushed a commit to Hikariii/kubernetes-vagrant-coreos-cluster that referenced this issue May 1, 2015
@pires
Desperate attempt at fixing so many issues that were brought by what …
7c45c7d 
…we thought were smart moves. Sorry for that!

Removing dockerized etcd 2.x
Removing dockerized flannel
Master node doesn't need to rely on its own dockercache for pulling images, for now.
Refs pires#46 pires#47 pires#32
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
3 participants
@AntonioMeireles @pires @ggarson