• Linux HA на основе Pacemaker
• Pacemaker, теория
• Example configurations

• Пакет Corosync

nodeN# apt install pacemaker

nodeN# crm_mon -1

node1# apt install crmsh

node1# crm status

node1# crm configure show

node1# crm configure

property dc-deadtime=120    # Для уменьшения вероятности split-brain (НЕ помогает)

property no-quorum-policy=ignore
primitive st-null stonith:null params hostlist="node1 node2"
clone fencing st-null

op timeout=240s             # Для долго стартующих ресурсов (перезапуск VM в KVM на резервном узле после падения основного)

show
commit
quit

nodeN# find /var/lib/pacemaker/cib/

nodeN# cat /var/lib/pacemaker/cib/cib.xml

Resource Agents

• crm_resource(8) - Linux man page

node1# crm ra classes

node1# crm ra list lsb

node1# crm ra list systemd

node1# crm ra list ocf heartbeat

node1# less /usr/lib/ocf/resource.d/heartbeat/IPaddr2

node1# OCF_ROOT=/usr/lib/ocf OCF_RESKEY_ip=192.168.X.10 OCF_RESKEY_nic=eth0 OCF_RESKEY_cidr_netmask=24 /usr/lib/ocf/resource.d/heartbeat/IPaddr2 start

node1# OCF_ROOT=/usr/lib/ocf OCF_RESKEY_ip=192.168.X.10 OCF_RESKEY_nic=eth0 OCF_RESKEY_cidr_netmask=24 /usr/lib/ocf/resource.d/heartbeat/IPaddr2 stop

node1# crm configure 
crm(live)configure# primitive pr_ftp lsb:proftpd
crm(live)configure# primitive pr_ip ocf:heartbeat:IPaddr2 params ip=192.168.X.10 cidr_netmask=32 nic=eth0
crm(live)configure# group gr_ftp_ip pr_ftp pr_ip

node1# export EDITOR=nano

crm(live)configure# edit pr_ip

crm(live)configure# edit gr_ftp_ip

crm(live)configure# show

crm(live)configure# commit
crm(live)configure# exit

node1# crm status

node1# crm resource
crm(live)resource# locate gr_ftp_ip
crm(live)resource# exit
или
node1# crm resource locate gr_ftp_ip

node1# crm configure show


node1# crm resource move gr_ftp_ip node2
или
node1# crm resource move pr_ftp node2


node1# crm configure show


node1# crm configure delete cli-prefer-gr_ftp_ip
или
node1# crm configure delete cli-prefer-pr_ftp
или
node1# crm resource unmigrate gr_ftp_ip
или
node1# crm resource unmigrate pr_ftp

node1# crm configure show

node1# crm resource stop gr_ftp_ip

node1# crm configure delete gr_ftp_ip pr_ftp pr_ip

# crm resource cleanup pr_fs_r0

# stonith_admin -c -H node2.corpX.un

pacemaker# crm_resource -C

# tail -f /var/log/pacemaker/pacemaker.log

node1:~# cibadmin -E --force

nodeN:~# init 6

• Set systemd service to execute after fstab mount
• Проблема воспроизводилась при сбойных сценариях с ресурсами, использующими Сервис OCFS2

# systemctl list-units --type=mount

...
  disk3.mount                   loaded active mounted /disk3
...

# systemctl edit pacemaker.service

[Unit]
After=disk3.mount
#After=disk2.mount disk3.mount

primitive st-ssh stonith:external/ssh params hostlist="node1 node2"
clone fencing st-ssh

# pacemaker 1.X ()
property stonith-enabled=false
property default-resource-stickiness=100