Remove etcd-proxy from all nodes and use etcd multiaccess

This commit is contained in:
Matthew Mosesohn
2016-11-09 13:31:12 +03:00
parent 6b0d26ddf0
commit 95b460ae94
23 changed files with 55 additions and 380 deletions

View File

@@ -6,21 +6,14 @@
- reload etcd
- wait for etcd up
- name: restart etcd-proxy
command: /bin/true
notify:
- etcd | reload systemd
- reload etcd-proxy
- wait for etcd up
- name: etcd | reload systemd
command: systemctl daemon-reload
when: ansible_service_mgr == "systemd"
- name: wait for etcd up
uri: url="http://{% if is_etcd_master %}{{ etcd_address }}{% else %}127.0.0.1{% endif %}:2379/health"
uri: url="http://{% if is_etcd_master %}{{ etcd_address }}{% else %}127.0.0.1{% endif %}:2379/health"
register: result
until: result.status == 200
until: result.status is defined and result.status == 200
retries: 10
delay: 5
@@ -29,9 +22,3 @@
name: etcd
state: restarted
when: is_etcd_master
- name: reload etcd-proxy
service:
name: etcd-proxy
state: restarted
when: is_etcd_proxy

View File

@@ -26,19 +26,3 @@
mode: 0755
when: ansible_service_mgr in ["sysvinit","upstart"] and ansible_os_family == "Debian" and is_etcd_master
notify: restart etcd
- name: Configure | Copy etcd-proxy.service systemd file
template:
src: "etcd-proxy-{{ etcd_deployment_type }}.service.j2"
dest: /etc/systemd/system/etcd-proxy.service
backup: yes
when: ansible_service_mgr == "systemd" and is_etcd_proxy
notify: restart etcd-proxy
- name: Configure | Write etcd-proxy initd script
template:
src: "deb-etcd-proxy-{{ etcd_deployment_type }}.initd.j2"
dest: /etc/init.d/etcd-proxy
owner: root
mode: 0755
when: ansible_service_mgr in ["sysvinit","upstart"] and ansible_os_family == "Debian" and is_etcd_proxy
notify: restart etcd-proxy

View File

@@ -1,8 +1,13 @@
---
- include: pre_upgrade.yml
- include: install.yml
when: is_etcd_master
- include: set_cluster_health.yml
when: is_etcd_master
- include: configure.yml
when: is_etcd_master
- include: refresh_config.yml
when: is_etcd_master
- name: Ensure etcd is running
service:
@@ -11,23 +16,11 @@
enabled: yes
when: is_etcd_master
- name: Ensure etcd-proxy is running
service:
name: etcd-proxy
state: started
enabled: yes
when: is_etcd_proxy
- name: Restart etcd if binary changed
command: /bin/true
notify: restart etcd
when: etcd_deployment_type == "host" and etcd_copy.stdout_lines and is_etcd_master
- name: Restart etcd-proxy if binary changed
command: /bin/true
notify: restart etcd-proxy
when: etcd_deployment_type == "host" and etcd_copy.stdout_lines and is_etcd_proxy
# Reload systemd before starting service
- meta: flush_handlers
@@ -35,4 +28,6 @@
# initial state of the cluster is in `existing`
# state insted of `new`.
- include: set_cluster_health.yml
when: is_etcd_master
- include: refresh_config.yml
when: is_etcd_master

View File

@@ -5,10 +5,3 @@
dest: /etc/etcd.env
notify: restart etcd
when: is_etcd_master
- name: Refresh config | Create etcd-proxy config file
template:
src: etcd-proxy.j2
dest: /etc/etcd-proxy.env
notify: restart etcd-proxy
when: is_etcd_proxy

View File

@@ -1,120 +0,0 @@
#!/bin/sh
set -a
### BEGIN INIT INFO
# Provides: etcd-proxy
# Required-Start: $local_fs $network $syslog
# Required-Stop:
# Default-Start: 2 3 4 5
# Default-Stop: 0 1 6
# Short-Description: etcd-proxy
# Description:
# etcd-proxy is a proxy for etcd: distributed, consistent key-value store for shared configuration and service discovery
### END INIT INFO
PATH=/sbin:/usr/sbin:/bin/:/usr/bin
DESC="etcd-proxy"
NAME=etcd-proxy
DAEMON={{ docker_bin_dir | default("/usr/bin") }}/docker
DAEMON_EXEC=`basename $DAEMON`
DAEMON_ARGS="run --restart=always --env-file=/etc/etcd-proxy.env \
--net=host \
--stop-signal=SIGKILL \
-v /usr/share/ca-certificates/:/etc/ssl/certs:ro \
--name={{ etcd_proxy_member_name | default("etcd-proxy") }} \
{{ etcd_image_repo }}:{{ etcd_image_tag }} \
{% if etcd_after_v3 %}
{{ etcd_container_bin_dir }}etcd
{% endif %}"
SCRIPTNAME=/etc/init.d/$NAME
DAEMON_USER=root
STOP_SCHEDULE="${STOP_SCHEDULE:-QUIT/5/TERM/5/KILL/5}"
PID=/var/run/etcd-proxy.pid
# Exit if the binary is not present
[ -x "$DAEMON" ] || exit 0
# Define LSB log_* functions.
# Depend on lsb-base (>= 3.2-14) to ensure that this file is present
# and status_of_proc is working.
. /lib/lsb/init-functions
do_status()
{
status_of_proc -p $PID "$DAEMON" "$NAME" && exit 0 || exit $?
}
# Function that starts the daemon/service
#
do_start()
{
{{ docker_bin_dir | default("/usr/bin") }}/docker rm -f {{ etcd_proxy_member_name | default("etcd-proxy") }} &>/dev/null || true
sleep 1
start-stop-daemon --background --start --quiet --make-pidfile --pidfile $PID --user $DAEMON_USER --exec $DAEMON -- \
$DAEMON_ARGS \
|| return 2
}
#
# Function that stops the daemon/service
#
do_stop()
{
start-stop-daemon --stop --quiet --retry=$STOP_SCHEDULE --pidfile $PID --name $DAEMON_EXEC
RETVAL="$?"
sleep 1
return "$RETVAL"
}
case "$1" in
start)
log_daemon_msg "Starting $DESC" "$NAME"
do_start
case "$?" in
0|1) log_end_msg 0 || exit 0 ;;
2) log_end_msg 1 || exit 1 ;;
esac
;;
stop)
log_daemon_msg "Stopping $DESC" "$NAME"
if do_stop; then
log_end_msg 0
else
log_failure_msg "Can't stop etcd-proxy"
log_end_msg 1
fi
;;
status)
if do_status; then
log_end_msg 0
else
log_failure_msg "etcd-proxy is not running"
log_end_msg 1
fi
;;
restart|force-reload)
log_daemon_msg "Restarting $DESC" "$NAME"
if do_stop; then
if do_start; then
log_end_msg 0
exit 0
else
rc="$?"
fi
else
rc="$?"
fi
log_failure_msg "Can't restart etcd-proxy"
log_end_msg ${rc}
;;
*)
echo "Usage: $SCRIPTNAME {start|stop|status|restart|force-reload}" >&2
exit 3
;;
esac

View File

@@ -1,110 +0,0 @@
#!/bin/sh
set -a
### BEGIN INIT INFO
# Provides: etcd-proxy
# Required-Start: $local_fs $network $syslog
# Required-Stop:
# Default-Start: 2 3 4 5
# Default-Stop: 0 1 6
# Short-Description: etcd-proxy
# Description:
# etcd-proxy is a proxy for etcd: distributed, consistent key-value store for shared configuration and service discovery
### END INIT INFO
PATH=/sbin:/usr/sbin:/bin:/usr/bin
DESC="etcd-proxy"
NAME=etcd-proxy
DAEMON={{ bin_dir }}/etcd
DAEMON_ARGS=""
SCRIPTNAME=/etc/init.d/$NAME
DAEMON_USER=etcd
STOP_SCHEDULE="${STOP_SCHEDULE:-QUIT/5/TERM/5/KILL/5}"
PID=/var/run/etcd-proxy.pid
# Exit if the binary is not present
[ -x "$DAEMON" ] || exit 0
# Read configuration variable file if it is present
[ -f /etc/etcd-proxy.env ] && . /etc/etcd-proxy.env
# Define LSB log_* functions.
# Depend on lsb-base (>= 3.2-14) to ensure that this file is present
# and status_of_proc is working.
. /lib/lsb/init-functions
do_status()
{
status_of_proc -p $PID "$DAEMON" "$NAME" && exit 0 || exit $?
}
# Function that starts the daemon/service
#
do_start()
{
start-stop-daemon --background --start --quiet --make-pidfile --pidfile $PID --user $DAEMON_USER --exec $DAEMON -- \
$DAEMON_ARGS \
|| return 2
}
#
# Function that stops the daemon/service
#
do_stop()
{
start-stop-daemon --stop --quiet --retry=$STOP_SCHEDULE --pidfile $PID --name $NAME
RETVAL="$?"
sleep 1
return "$RETVAL"
}
case "$1" in
start)
log_daemon_msg "Starting $DESC" "$NAME"
do_start
case "$?" in
0|1) log_end_msg 0 || exit 0 ;;
2) log_end_msg 1 || exit 1 ;;
esac
;;
stop)
log_daemon_msg "Stopping $DESC" "$NAME"
if do_stop; then
log_end_msg 0
else
log_failure_msg "Can't stop etcd-proxy"
log_end_msg 1
fi
;;
status)
if do_status; then
log_end_msg 0
else
log_failure_msg "etcd-proxy is not running"
log_end_msg 1
fi
;;
restart|force-reload)
log_daemon_msg "Restarting $DESC" "$NAME"
if do_stop; then
if do_start; then
log_end_msg 0
exit 0
else
rc="$?"
fi
else
rc="$?"
fi
log_failure_msg "Can't restart etcd-proxy"
log_end_msg ${rc}
;;
*)
echo "Usage: $SCRIPTNAME {start|stop|status|restart|force-reload}" >&2
exit 3
;;
esac

View File

@@ -1,28 +0,0 @@
[Unit]
Description=etcd-proxy docker wrapper
Wants=docker.socket
After=docker.service
[Service]
User=root
PermissionsStartOnly=true
ExecStart={{ docker_bin_dir | default("/usr/bin") }}/docker run --restart=always \
--env-file=/etc/etcd-proxy.env \
{# TODO(mattymo): Allow docker IP binding and disable in envfile
-p 2380:2380 -p 2379:2379 #}
--net=host \
--stop-signal=SIGKILL \
-v /usr/share/ca-certificates/:/etc/ssl/certs:ro \
--name={{ etcd_proxy_member_name | default("etcd-proxy") }} \
{{ etcd_image_repo }}:{{ etcd_image_tag }} \
{% if etcd_after_v3 %}
{{ etcd_container_bin_dir }}etcd
{% endif %}
ExecStartPre=-{{ docker_bin_dir | default("/usr/bin") }}/docker rm -f {{ etcd_proxy_member_name | default("etcd-proxy") }}
ExecReload={{ docker_bin_dir | default("/usr/bin") }}/docker restart {{ etcd_proxy_member_name | default("etcd-proxy") }}
ExecStop={{ docker_bin_dir | default("/usr/bin") }}/docker stop {{ etcd_proxy_member_name | default("etcd-proxy") }}
Restart=always
RestartSec=15s
[Install]
WantedBy=multi-user.target

View File

@@ -1,19 +0,0 @@
[Unit]
Description=etcd-proxy
After=network.target
[Service]
Type=notify
User=etcd
PermissionsStartOnly=true
EnvironmentFile=/etc/etcd-proxy.env
ExecStart={{ bin_dir }}/etcd
ExecStartPre=/bin/mkdir -p /var/lib/etcd-proxy
ExecStartPre=/bin/chown -R etcd: /var/lib/etcd-proxy
NotifyAccess=all
Restart=always
RestartSec=10s
LimitNOFILE=40000
[Install]
WantedBy=multi-user.target

View File

@@ -1,5 +0,0 @@
ETCD_DATA_DIR=/var/lib/etcd-proxy
ETCD_PROXY=on
ETCD_LISTEN_CLIENT_URLS={{ etcd_access_endpoint }}
ETCD_NAME={{ etcd_proxy_member_name | default("etcd-proxy") }}
ETCD_INITIAL_CLUSTER={{ etcd_peer_addresses }}

View File

@@ -3,11 +3,7 @@ ETCD_ADVERTISE_CLIENT_URLS={{ etcd_client_url }}
ETCD_INITIAL_ADVERTISE_PEER_URLS={{ etcd_peer_url }}
ETCD_INITIAL_CLUSTER_STATE={% if etcd_cluster_is_healthy.rc != 0 | bool %}new{% else %}existing{% endif %}
{% if not is_etcd_proxy %}
ETCD_LISTEN_CLIENT_URLS=http://{{ etcd_address }}:2379,http://127.0.0.1:2379
{% else %}
ETCD_LISTEN_CLIENT_URLS=http://{{ etcd_address }}:2379
{% endif %}
ETCD_ELECTION_TIMEOUT=10000
ETCD_INITIAL_CLUSTER_TOKEN=k8s_etcd
ETCD_LISTEN_PEER_URLS=http://{{ etcd_address }}:2380

View File

@@ -26,7 +26,7 @@ spec:
image: calico/kube-policy-controller:latest
env:
- name: ETCD_ENDPOINTS
value: "{{ etcd_endpoint }}"
value: "{{ etcd_access_endpoint }}"
# Location of the Kubernetes API - this shouldn't need to be
# changed so long as it is used in conjunction with
# CONFIGURE_ETC_HOSTS="true".
@@ -38,3 +38,12 @@ spec:
# This removes the need for KubeDNS to resolve the Service.
- name: CONFIGURE_ETC_HOSTS
value: "true"
volumeMounts:
- mountPath: {{ etcd_cert_dir }}
name: etcd-certs
readOnly: true
volumes:
- hostPath:
path: {{ etcd_cert_dir }}
name: etcd-certs

View File

@@ -14,12 +14,3 @@
name: kube-apiserver
state: stopped
when: (kube_apiserver_service_file.stat.exists|default(False) or kube_apiserver_init_script.stat.exists|default(False))
- name: "Pre-upgrade | remove kube-apiserver service definition"
file:
path: "{{ item }}"
state: absent
when: (kube_apiserver_service_file.stat.exists|default(False) or kube_apiserver_init_script.stat.exists|default(False))
with_items:
- /etc/systemd/system/kube-apiserver.service
- /etc/init.d/kube-apiserver

View File

@@ -62,3 +62,4 @@ spec:
- hostPath:
path: /var/log/
name: logfile

View File

@@ -1,6 +1,7 @@
{
"name": "calico-k8s-network",
"type": "calico",
"etcd_endpoints": "{{ etcd_access_endpoint }}",
"log_level": "info",
"ipam": {
"type": "calico-ipam"

View File

@@ -45,3 +45,6 @@ openstack_username: "{{ lookup('env','OS_USERNAME') }}"
openstack_password: "{{ lookup('env','OS_PASSWORD') }}"
openstack_region: "{{ lookup('env','OS_REGION_NAME') }}"
openstack_tenant_id: "{{ lookup('env','OS_TENANT_ID') }}"
# All clients access each node individually, instead of using a load balancer.
etcd_multiaccess: true

View File

@@ -30,7 +30,7 @@
- set_fact:
etcd_access_addresses: |-
{% for item in groups['etcd'] -%}
http://{{ hostvars[item].access_ip | default(hostvars[item].ip | default(hostvars[item].ansible_default_ipv4['address'])) }}:2379{% if not loop.last %},{% endif %}
http://{{ item }}:2379{% if not loop.last %},{% endif %}
{%- endfor %}
- set_fact: etcd_access_endpoint="{% if etcd_multiaccess %}{{ etcd_access_addresses }}{% else %}{{ etcd_endpoint }}{% endif %}"
- set_fact:
@@ -43,13 +43,6 @@
{% for item in groups['etcd'] -%}
{{ "etcd"+loop.index|string }}=http://{{ hostvars[item].access_ip | default(hostvars[item].ip | default(hostvars[item].ansible_default_ipv4['address'])) }}:2380{% if not loop.last %},{% endif %}
{%- endfor %}
- set_fact:
etcd_proxy_member_name: |-
{% for host in groups['k8s-cluster'] %}
{% if inventory_hostname == host %}{{"etcd-proxy"+loop.index|string }}{% endif %}
{% endfor %}
- set_fact:
is_etcd_proxy: "{{ inventory_hostname in groups['k8s-cluster'] }}"
- set_fact:
is_etcd_master: "{{ inventory_hostname in groups['etcd'] }}"
- set_fact:

View File

@@ -43,17 +43,21 @@
- name: Calico | wait for etcd
uri: url=http://localhost:2379/health
register: result
until: result.status == 200
until: result.status == 200 or result.status == 401
retries: 10
delay: 5
when: inventory_hostname in groups['kube-master']
delegate_to: "{{groups['etcd'][0]}}"
run_once: true
- name: Calico | Check if calico network pool has already been configured
uri:
url: "{{ etcd_endpoint }}/v2/keys/calico/v1/ipam/v4/pool"
return_content: yes
status_code: 200,404
command: |-
curl \
--cacert {{ etcd_cert_dir }}/ca.pem \
--cert {{ etcd_cert_dir}}/admin.pem \
--key {{ etcd_cert_dir }}/admin-key.pem \
https://localhost:2379/v2/keys/calico/v1/ipam/v4/pool
register: calico_conf
delegate_to: "{{groups['etcd'][0]}}"
run_once: true
- name: Calico | Define ipip pool argument
@@ -79,21 +83,25 @@
environment:
NO_DEFAULT_POOLS: true
run_once: true
when: calico_conf.status == 404 or "nodes" not in calico_conf.content
when: '"Key not found" in calico_conf.stdout or "nodes" not in calico_conf.stdout'
- name: Calico | Get calico configuration from etcd
uri:
url: "{{ etcd_endpoint }}/v2/keys/calico/v1/ipam/v4/pool"
return_content: yes
register: calico_pools
command: |-
curl http://localhost:2379/v2/keys/calico/v1/ipam/v4/pool
register: calico_pools_raw
delegate_to: "{{groups['etcd'][0]}}"
run_once: true
- set_fact:
calico_pools: "{{ calico_pools_raw.stdout | from_json }}"
run_once: true
- name: Calico | Check if calico pool is properly configured
fail:
msg: 'Only one network pool must be configured and it must be the subnet {{ kube_pods_subnet }}.
Please erase calico configuration and run the playbook again ("etcdctl rm --recursive /calico/v1/ipam/v4/pool")'
when: ( calico_pools.json['node']['nodes'] | length > 1 ) or
( not calico_pools.json['node']['nodes'][0]['key'] | search(".*{{ kube_pods_subnet | ipaddr('network') }}.*") )
when: ( calico_pools['node']['nodes'] | length > 1 ) or
( not calico_pools['node']['nodes'][0]['key'] | search(".*{{ kube_pods_subnet | ipaddr('network') }}.*") )
run_once: true
- name: Calico | Write /etc/network-environment
@@ -131,4 +139,3 @@
shell: "{{ bin_dir }}/calicoctl node bgp peer add {{ item.router_id }} as {{ item.as }}"
with_items: peers
when: peer_with_router|default(false) and inventory_hostname in groups['kube-node']

View File

@@ -1,8 +1,8 @@
[Unit]
Description=Calico per-node agent
Documentation=https://github.com/projectcalico/calico-docker
After=docker.service docker.socket etcd-proxy.service
Wants=docker.socket etcd-proxy.service
After=docker.service docker.socket
Wants=docker.socket
[Service]
User=root

View File

@@ -1,6 +1,7 @@
#!/bin/bash
/usr/bin/docker run --privileged --rm \
--net=host --pid=host -e ETCD_AUTHORITY={{ etcd_authority }} \
--net=host --pid=host \
-e ETCD_ENDPOINTS={{ etcd_access_endpoint }} \
-v /usr/bin/docker:/usr/bin/docker \
-v /var/run/docker.sock:/var/run/docker.sock \
-v /var/run/calico:/var/run/calico \

View File

@@ -3,7 +3,7 @@
DEFAULT_IPV4={{ip | default(ansible_default_ipv4.address) }}
# The Kubernetes master IP
KUBERNETES_MASTER={{ first_kube_master }}
KUBERNETES_MASTER={{ kube_apiserver_endpoint }}
# IP and port of etcd instance used by Calico
ETCD_AUTHORITY={{ etcd_authority }}
ETCD_ENDPOINTS={{ etcd_access_endpoint }}

View File

@@ -21,7 +21,7 @@
args:
- "--network-config=/etc/flannel-network.json"
- "--etcd-prefix=/{{ cluster_name }}/network"
- "--etcd-server={{ etcd_endpoint }}"
- "--etcd-endpoints={{ etcd_access_endpoint }}"
volumeMounts:
- name: "networkconfig"
mountPath: "/etc/flannel-network.json"