Initial add of docker-multinode by luxas · Pull Request #12 · kubernetes-retired/kube-deploy

luxas · 2016-05-03T20:44:21Z

Here's kubernetes/kubernetes#21646 for this repo instead.
Ref: #10

The scripts here are refactored, but this hasn't been merged into Kuberenetes core, so we should review it and test it here.
I'm sure there are bugs around (haven't touched this for about two months), so feel free to comment on it.

The docs around this are just unorganized, mostly outdated and kind of ugly.
We should link to guides in the docs for example instead of having a testing section here
We should discuss how this updates should be made to the docs repo when updating something here etc.

It should be much easier to run this guide too, and I'm proposing to have just one script (we may call it kube-up or whatever) instead of many small (master, worker, etc)
This doesn't work with docker-1.11
We should use @ArtfulCoder's one-container DNS solution instead of the current one I think

And the long-term goal would probably to be to write this purely in Go instead.

I will update this when I have time, and when we have an initial draft, let's merge it.

@johndmulhausen @mikedanese @vishh @janetkuo @fgrzadkowski @cheld @zreigz @batikanu @jimmyjones2 @jumpkick @yidinghan @brendandburns

cheld · 2016-05-04T06:40:14Z

+KUBE_ROOT=$(dirname "${BASH_SOURCE}")/../../..
+
+# Source useful scripts
+source "${KUBE_ROOT}/hack/lib/util.sh"


not working any more - copy them over?

I have errors:

./common.sh: line 48: ./../../../hack/lib/util.sh: No such file or directory ./common.sh: line 49: ./../../../cluster/lib/logging.sh: No such file or directory ./common.sh: line 52: kube::log::install_errexit: command not found ./common.sh: line 65: kube::log::error: command not found

zreigz · 2016-05-04T06:50:43Z

+      --iface="${FLANNEL_IFACE}"
+
+  # Wait for the flannel subnet.env file to be created instead of a timeout. This is faster and more reliable
+  local FLANNEL_SECONDS=0


You can use $SECONDS as an internal shell variable.
It is automatically increased after second. You just need reset this variable before while loop: SECONDS=0

cheld · 2016-05-04T09:06:00Z

Very nice. I am done with review.

PTAL

cheld · 2016-05-04T14:45:19Z

+Now you can list the nodes:
+
+```sh
+kubectl get nodes


cheld · 2016-05-04T14:49:38Z

One more question. In this docker provisioning the master is also minion. This is not the case in salt.

What is the advantage of this? Is the intention to include the master in the daemon sets?

jimmyjones2 · 2016-05-04T20:08:59Z

@luxas v1.2.0-alpha.7 is hardcoded in a number of places. Should this at least be updated to 1.2.0 - there is no reason anyone would want to use the alpha release?

jimmyjones2 · 2016-05-04T20:09:26Z

+FLANNEL_VERSION=${FLANNEL_VERSION:-"0.5.5"}
+FLANNEL_IFACE=${FLANNEL_IFACE:-"eth0"}
+FLANNEL_IPMASQ=${FLANNEL_IPMASQ:-"true"}
+FLANNEL_BACKEND=${FLANNEL_BACKEND:-"vxlan"}


Default to UDP for maximum compatibility? According to this vxlan is broken on CentOS 6, for example

As far as I remember, Docker is not fully working in CentOS. I would leave it as it is

jimmyjones2 · 2016-05-04T20:10:05Z

@luxas Should these scripts also starts cluster addons like DNS, UI etc, or at least have an option to?

jimmyjones2 · 2016-05-04T20:11:43Z

+      --cluster-dns=${DNS_SERVER_IP} \
+      --cluster-domain=${DNS_DOMAIN} \
+      --containerized \
+      --v=2


The guide also has --address=0.0.0.0 --enable-server Which is right?
Also needs -v /var/log/containers/:/var/log/containers:rw to support logging

luxas · 2016-05-05T14:11:51Z

Second rev pushed.
This is huge, please comment more, and it takes some time to get right.
You may see what I did now by reviewing the second commit.

jimmyjones2 · 2016-05-05T20:44:52Z

 ## Prerequisites

-The only thing you need is a machine with **Docker 1.7.1 or higher**
+The only thing you need is a machine with **Docker 1.8.3 or higher**


Why is 1.8.3 now necessary? It is just for docker daemon vs docker -d? 1.7.1 is the newest version CentOS 6 will ever get, and I got the previous scripts to successfully work with it.

Kubernetes officially supports 1.8.3-1.10.3 with release 1.2.
Sorry for CentOS 6, but it is really worth it to not upgrade?

:( Fair enough. Sadly plenty of corporate luttides out there!

cheld · 2016-05-09T12:45:05Z

2 small comments. PTAL

luxas · 2016-05-10T19:23:29Z

Third commit pushed.
PTAL @cheld and others

cheld · 2016-05-11T03:32:39Z

+      --data-dir=/var/etcd/data
+
+  # Wait for etcd to come up
+  sleep 5


@batikanu you investigated more time in analysis of the curl on etcd. Better idea?

zreigz · 2016-05-12T08:25:57Z

I've made some effort to be able test this solution automatically. I've prepared Vagrant project which can create VM instances for master and worker. Additionally it:

install docker
clone https://github.com/luxas/kube-deploy.git (later kubernetes branch)
execute master.sh/node.sh

You can find this here: https://github.com/FujitsuEnablingSoftwareTechnologyGmbH/kube-deploy-vagrant

It uses centos 7 image with the newest Docker 1.11.1. Unfortunately the newest docker has bug and cannot start images from “bootstrap” instance of the Docker daemon. I've created bug for this:
moby/moby#22684

I will try downgrade docker and continue testing. With this vagrant cluster I will be able run e2e tests.

zreigz · 2016-05-12T11:18:30Z

I've managed downgrade docker to 1.10.3 and master started correctly. I noticed problem with worker on node instance. It seems to started correctly but kubectl get nodes doesn't show this worker. When I added
--hostname-override=127.0.0.1 after this line https://github.com/luxas/kube-deploy/blob/docker_multinode/docker-multinode/common.sh#L331
then worker was visible. Any idea ?

googlebot added the cla: yes label May 3, 2016

cheld reviewed May 4, 2016
View reviewed changes

mikedanese self-assigned this May 4, 2016

zreigz reviewed May 4, 2016
View reviewed changes

cheld reviewed May 4, 2016
View reviewed changes

Comment thread docker-multinode/master.md Outdated

Now you can list the nodes:

```sh

kubectl get nodes

cheld May 4, 2016

Copy link
Copy Markdown

Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

$ missing

jimmyjones2 reviewed May 4, 2016
View reviewed changes

luxas mentioned this pull request May 5, 2016

implement v2 of kubernetes deployments kubernetes/kubernetes#23174

Closed

jimmyjones2 reviewed May 5, 2016
View reviewed changes

cheld reviewed May 11, 2016
View reviewed changes

Conversation

luxas commented May 3, 2016 • edited by mikedanese Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

cheld commented May 4, 2016

Uh oh!

Choose a reason for hiding this comment

Uh oh!

cheld commented May 4, 2016

Uh oh!

jimmyjones2 commented May 4, 2016

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

jimmyjones2 commented May 4, 2016

Uh oh!

Choose a reason for hiding this comment

Uh oh!

luxas commented May 5, 2016

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

cheld commented May 9, 2016

Uh oh!

luxas commented May 10, 2016

Uh oh!

Choose a reason for hiding this comment

Uh oh!

zreigz commented May 12, 2016

Uh oh!

zreigz commented May 12, 2016

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

7 participants

luxas commented May 3, 2016 •

edited by mikedanese

Loading