x
/
gluon
Code Issues Proposed changes
gluon/doc/source/installation/install_etcd.rst

6.6 KiB
Raw Blame History

Install OpenStack

Before you start installing etcd, please make sure that you have installed OpenStack Pike Release. For example, to install devstack Pike release:

$ git clone https://github.com/openstack-dev/devstack -b stable/pike

# Then generate local.conf in devstack directory, then run
$ ./stack.sh

Stop etcd3 in Devstack Pike Release

OpenStack Pike Release starts etcd3 by default when user starts stack, while Gluon is based on etcd-v2.3.6. This will cause issue for Gluon. Thus you need to stop etcd3 that is started by OpenStack Pike.

For Devstack Pike Release:

$ sudo systemctl stop devstack@etcd.service

You may double check to make sure there is no etcd3 running:

$ ps -aef | grep etcd

You should not see any etcd process running.

Install etcd for Gluon

The following instructions are from reference1.

You need at least three nodes for the etcd cluster to work, e.g. one controller and two computes.

On Each Node

STEP-1: Download etcd Package

$ curl -L  https://github.com/coreos/etcd/releases/download/v2.3.6/etcd-v2.3.6-linux-amd64.tar.gz -o etcd-v2.3.6-linux-amd64.tar.gz

# Then Unzip/Untar the downloaded file

STEP-2: Copy executables to /usr/local/bin

$ cd etcd-v2.3.6-linux-amd64
$ sudo cp etcd /usr/local/bin
$ sudo cp etcdctl /usr/local/bin

STEP-3: Create a directory for etcd data

$ sudo mkdir /var/etcd

STEP-4: Create upstart init file:

In /etc/init directory, create a file called etcd.conf, and paste the following into it:

description "etcd 2.0 distributed key-value store"
author "Scott Lowe <scott.lowe@scottlowe.org>"
start on (net-device-up
          and local-filesystems
          and runlevel [2345])
stop on runlevel [016]
respawn
respawn limit 10 5
script
  if [ -f "/etc/default/etcd" ]; then
    . /etc/default/etcd
  fi
chdir /var/etcd
exec /usr/local/bin/etcd >>/var/log/etcd.log 2>&1
end script

STEP-5: Create an override file for etcd parameters:

In /etc/init directory, create a file called etcd.override and paste the following into it:

# Override file for etcd Upstart script providing some environment variables
env ETCD_INITIAL_CLUSTER="etcd-01=http://10.2.0.32:2380,etcd-02=http://10.2.0.102:2380,etcd-03=http://10.2.0.101:2380"
env ETCD_INITIAL_CLUSTER_STATE="new"
env ETCD_INITIAL_CLUSTER_TOKEN="etcd-cluster-1"
env ETCD_INITIAL_ADVERTISE_PEER_URLS="http://10.2.0.32:2380"
env ETCD_DATA_DIR="/var/etcd"
env ETCD_LISTEN_PEER_URLS="http://10.2.0.32:2380"
env ETCD_LISTEN_CLIENT_URLS="http://10.2.0.32:2379,http://127.0.0.1:2379"
env ETCD_ADVERTISE_CLIENT_URLS="http://10.2.0.32:2379"
env ETCD_NAME="etcd-01"

NOTE:

  • The IP Addresses will need to be changed for your own machines!
  • For each node in the cluster, the file will be slightly different, i.e. the IP address of "advertise" and "listen" URLs, and etcd names will be for each specific node.

For instance, the files on the other two nodes would look like:

# Override file for etcd Upstart script providing some environment variables
env ETCD_INITIAL_CLUSTER="etcd-01=http://10.2.0.32:2380,etcd-02=http://10.2.0.102:2380,etcd-03=http://10.2.0.101:2380"
env ETCD_INITIAL_CLUSTER_STATE="new"
env ETCD_INITIAL_CLUSTER_TOKEN="etcd-cluster-1"
env ETCD_INITIAL_ADVERTISE_PEER_URLS="http://10.2.0.102:2380"
env ETCD_DATA_DIR="/var/etcd"
env ETCD_LISTEN_PEER_URLS="http://10.2.0.102:2380"
env ETCD_LISTEN_CLIENT_URLS="http://10.2.0.102:2379,http://127.0.0.1:2379"
env ETCD_ADVERTISE_CLIENT_URLS="http://10.2.0.102:2379"
env ETCD_NAME="etcd-02"
# Override file for etcd Upstart script providing some environment variables
env ETCD_INITIAL_CLUSTER="etcd-01=http://10.2.0.32:2380,etcd-02=http://10.2.0.102:2380,etcd-03=http://10.2.0.101:2380"
env ETCD_INITIAL_CLUSTER_STATE="new"
env ETCD_INITIAL_CLUSTER_TOKEN="etcd-cluster-1"
env ETCD_INITIAL_ADVERTISE_PEER_URLS="http://10.2.0.101:2380"
env ETCD_DATA_DIR="/var/etcd"
env ETCD_LISTEN_PEER_URLS="http://10.2.0.101:2380"
env ETCD_LISTEN_CLIENT_URLS="http://10.2.0.101:2379,http://127.0.0.1:2379"
env ETCD_ADVERTISE_CLIENT_URLS="http://10.2.0.101:2379"
env ETCD_NAME="etcd-03"

STEP-6: Adjust iptables:

$ sudo iptables -A INPUT -p tcp -m multiport --ports 2380,2379 -m comment --comment "etcd" -j ACCEPT
$ sudo invoke-rc.d iptables-persistent save

# Note: for Ubuntu 16.04, you may have to use netfilter-persistent as follows:
# sudo apt-get install netfilter-persistent
# sudo invoke-rc.d netfilter-persistent save

STEP-7: Start the etcd server:

As root:

$ initctl start etcd

Or on ubuntu 14.04, run:

$ sudo start etcd

STEP-8: Verify the cluster is healthy:

$ etcdctl cluster-health
member 5cd8baf7fb9d49b7 is healthy: got healthy result from http://10.2.0.102:2379
member 9e95400273fd2acb is healthy: got healthy result from http://10.2.0.101:2379
member ce8a4cd91a34b3f2 is healthy: got healthy result from http://10.2.0.32:2379
cluster is healthy

References

  1. http://blog.scottlowe.org/2015/04/15/running-etcd-20-cluster/↩︎