ipv6 address configuration
The ipv6 address of the cloud host needs to be applied to the IT department for ipv6, and they will assign you the address and gateway.
If you already have an ipv6 address and gateway, set it up as follows.
Locate the configuration sysctl.conf file at the path: /etc/sysctl.conf and add/modify the following configuration.
net.ipv6.conf.all.disable_ipv6 = 0
net.ipv6.conf.default.disable_ipv6 = 0
net.ipv6.conf.lo.disable_ipv6 = 0
Then execute the sysctl -p command to make the above changes take effect
Find the network.conf file at /etc/sysconfig/network and modify/add the following configuration
NETWORKING_IPV6=yes
Configure the IPV6 address.
vi /etc/sysconfig/network-scripts/ifcfg-eth1 #eth0 or eth1 depending on the specific NIC of the machine.
# Created by cloud-init on instance boot automatically, do not edit.
#
BOOTPROTO=none
DEVICE=eth1 #NIC
MTU=1500
ONBOOT=yes
STARTMODE=auto
TYPE=Ethernet
USERCTL=no
IPV6_AUTOCONF=no
IPV6_DEFROUTE=yes
IPV6_FAILURE_FATAL=no
IPV6INIT=yes
IPV6ADDR=FD69:1:0010:F940::A13:24CE/64 #ipv6 address
IPV6_DEFAULTGW=FD69:1:0010:F940::A13:24FE #gateway
#Reboot the NIC
systemctl restart network
After applying for the address, configure it to ensure that it can be pinged directly to each other and ssh can log in, e.g.
ping6 -I eth1 fd69:1:10:f940::a13:24ce
ssh adbpuq@fd69:1:10:f940::a13:24ce -p 22022
Main library configuration
The same as the above steps to build the host, you need to pay attention to the configuration file modification.
#postgresql.conf
listen_addresses = '*'
#pg_hba.conf
# add under IPv6 local connections:
host all all ::/0 trust
host replication all ::/0 trust
etcd configuration
etcd is set up as above, only the ip section of etcd's configuration file /etc/etcd/conf.yaml needs to be changed to ipv6 address, and it should be noted that the ipv6 address needs to be enclosed in [ ].
As an example, the configuration of one machine, /etc/etcd/conf.yaml, is as follows
name: etcd-2
data-dir: /var/lib/etcd/data
listen-client-urls: http://[fd69:1:10:f940::a13:24ce]:12379,http://[::1]:12379
advertise-client-urls: http://[fd69:1:10:f940::a13:24ce]:12379
listen-peer-urls: http://[fd69:1:10:f940::a13:24ce]:12380
initial-advertise-peer-urls: http://[fd69:1:10:f940::a13:24ce]:12380
initial-cluster: etcd-1=http://[fd69:1:10:f780::a13:1c81]:12380,etcd-2=http://[fd69:1:10:f940::a13:24ce]:12380,etcd-3=http://[fd69:1:10:f940::a13:24cf]:12380
initial-cluster-token: etcd-cluster-token
initial-cluster-state: existing
heartbeat-interval: 1000
election-timeout: 5000
etcd configurationbasic authentication
This step only needs to be run on the leader (just run it on one machine) and etcd will automatically synchronize.
#fd69:1:10:f780::a13:1c81，fd69:1:10:f940::a13:24ce，fd69:1:10:f940::a13:24cf]:12379 are ipv6 for three machines, which need to be modified to suit the environment.
# antdb user execution
# Enable v3 privilege authentication
export ETCDCTL_API=3
etcdctl --endpoints=http://[fd69:1:10:f780::a13:1c81]:12379,http://[fd69:1:10:f940::a13:24ce]:12379,http://[fd69:1:10:f940::a13:24cf]:12379 user add root
--Follow the prompts to enter the password (set your own password, e.g. antdb)
etcdctl --endpoints=http://[fd69:1:10:f780::a13:1c81]:12379,http://[fd69:1:10:f940::a13:24ce]:12379,http://[fd69:1:10:f940::a13:24cf]:12379 auth enable
# Enable v2 privilege authentication
export ETCDCTL_API=2
etcdctl --endpoints=http://[fd69:1:10:f780::a13:1c81]:12379,http://[fd69:1:10:f940::a13:24ce]:12379,http://[fd69:1:10:f940::a13:24cf]:12379 user passwd root
--Follow the prompts to enter your password (set your own password:, e.g. antdb)
# Create a patroni user and set a password
etcdctl --endpoints=http://[fd69:1:10:f780::a13:1c81]:12379,http://[fd69:1:10:f940::a13:24ce]:12379,http://[fd69:1:10:f940::a13:24cf]:12379 --username root:antdb role add patroni_user
# -path '/service/*' This is the content of the key of etcd, which looks like a path, in fact, it is a string, in order to distinguish the hierarchical relationship, the key is generally defined as a directory, you can later match the prefix to find. It is consistent with the namespace: /service/ setting in /etc/patroni.yml.
etcdctl --endpoints=http://[fd69:1:10:f780::a13:1c81]:12379,http://[fd69:1:10:f940::a13:24ce]:12379,http://[fd69:1:10:f940::a13:24cf]:12379 --username root:antdb role grant patroni_user -path '/service/*' -readwrite
etcdctl --endpoints=http://[fd69:1:10:f780::a13:1c81]:12379,http://[fd69:1:10:f940::a13:24ce]:12379,http://[fd69:1:10:f940::a13:24cf]:12379 --username root:antdb user add patroni_user
etcdctl --endpoints=http://[fd69:1:10:f780::a13:1c81]:12379,http://[fd69:1:10:f940::a13:24ce]:12379,http://[fd69:1:10:f940::a13:24cf]:12379 --username root:antdb user grant patroni_user -roles patroni_user
#View:
etcdctl --endpoints=http://[fd69:1:10:f780::a13:1c81]:12379,http://[fd69:1:10:f940::a13:24ce]:12379,http://[fd69:1:10:f940::a13:24cf]:12379 --username root:antdb role list
patroni configuration
The patroni configuration is the same as before, just replace the original ipv4 part with ipv6. Take one of the machines as an example.
scope: postgres-cluster
namespace: /service/
name: pgnode03
restapi:
listen: fd69:1:10:f940::a13:24cf:8008
connect_address: fd69:1:10:f940::a13:24cf:8008
etcd:
hosts: fd69:1:10:f780::a13:1c81:12379,fd69:1:10:f940::a13:24ce:12379,fd69:1:10:f940::a13:24cf:12379
username: patroni_user
password: antdb
bootstrap:
# this section will be written into Etcd:/<namespace>/<scope>/config after initializing new cluster
# and all other cluster members will use it as a `global configuration`
dcs:
ttl: 15
loop_wait: 5
retry_timeout: 5
maximum_lag_on_failover: 1048576
master_start_timeout: 300
synchronous_mode: true
postgresql:
use_pg_rewind: true
parameters:
listen_addresses: "*"
port: 55551
wal_level: logical
log_directory: "pg_log"
log_destination: "csvlog"
hot_standby: "on"
wal_keep_size: 10240
max_wal_senders: 10
postgresql:
listen: fd69:1:10:f940::a13:24cf:55551
connect_address: fd69:1:10:f940::a13:24cf:55551
data_dir: /data1/adbpuq/data
bin_dir: /data1/adbpuq/pgsql/bin
pgpass: /data1/adbpuq/.pgpass
authentication:
replication:
username: replicator
password: antdb
superuser:
username: adbpuq
password: AntDB@2022@04
parameters:
unix_socket_directories: '/tmp'
tags:
nofailover: false
noloadbalance: false
clonefrom: false
nosync: false
The above steps are the parts of ipv6 that distinguish ipv4 from ipv6. The rest of the operation is inconvenient.
Problems during patroni installation
- Error "failed import ......" is reported
patroni depends on many installation packages, you can use the pip command to install them in order according to the error message.
For example:
pip3 install python-consul2
pip3 install python-etcd
pip3 install dnspython
pip3 install kazoo
2, /etc/patroni.yml file format problems
yml format document has some formatting requirements: use indentation to indicate hierarchical relationships; do not allow the use of tabs, only spaces are allowed.
3、Etcd: failed to send out heartbeat. Heartbeat check is not allowed
yum install ntp
systemctl start ntpd
Top comments (0)