Zabbix cluster using Pcs & Crosynce

[batchenr]

Hello,

i just made some tets on Pcs & Crosynce cluster on zabbix server
and it all worked fine, then i did the same on the prod server (ofcurse i backed up)

and then i started to get errors.
now, i have 2 zabbix servers 3.2.6 Centos 7 + mariadb
and 1 remote centos 7 maria db

all works fine with 1 server and remote db.
when i did the cluster install it pass but i got errors and it didnt worked correctly.

1. first of all i made a virtual ip and it works fine, but i saw in the zabbix_agent log that the virtual ip is not the one trying to connect to the agent, but the internal node1 node2 servers - why ?

i want to manage it from 1 ip.

2. when i did a test and rebooted node1 - zabbix server on node 2 got DB error

"query failed: [1062] Duplicate entry '2059346' for key 'PRIMARY' [insert into events (eventid,source,object,objectid,clock,ns,value) values"

how can i overcome duplicates ?

other then that it works fine

[Akansha123]

As you said you are getting node1 and node2 IP instead of virtualIP. In this case you have to change "Server" and "Server_Active" parameter with <cluster_IP> or <virtual_IP> in zabbix_agentd.conf file

Through corosync & pacemaker what type of high availability you are trying to achieve? As below three possibilities exist
1) HA of zabbix server
2) HA of database
3) HA of GUI

Are you using "HA of database" with shared database on 3rd node? Then you wont' be requiring separate databases on node 1 and node 2.

You are using which type of configuration Active-Active HA or Active-Passive HA?

[batchenr]

As you said you are getting node1 and node2 IP instead of virtualIP. In this case you have to change "Server" and "Server_Active" parameter with <cluster_IP> or <virtual_IP> in zabbix_agentd.conf file

Through corosync & pacemaker what type of high availability you are trying to achieve? As below three possibilities exist
1) HA of zabbix server
2) HA of database
3) HA of GUI

Are you using "HA of database" with shared database on 3rd node? Then you wont' be requiring separate databases on node 1 and node 2.

You are using which type of configuration Active-Active HA or Active-Passive HA?

Hi,
thanks for the replay.

when i changed the Server & Server active at the agent conf i could see
at agent.log that it saying premmition denaind to node1&node2
and only virtual IP allowd - but the virtual ip is not the one who is making the connection.

maybe it is related to zabbix.conf settings :
SourceIP
ListenIP

?

and for now i want a- passiv active cluster if im correct
i want 2 zabbix servers machine that only one is active and when the main machine is powerd off \ zabbix crashes the other machine takes over.

and i have for now one remote db server and it the futuer i would like to connect it to a cluster also - passiv active.

now is i do a cluster status this is what is shows :

[root@Cluster-Tester-01 ~]# pcs status
Cluster name: zabbixserver
WARNING: corosync and pacemaker node names do not match (IPs used in setup?)
Stack: corosync
Current DC: Cluster-tester-02 (version 1.1.15-11.el7_3.4-e174ec8) - partition with quorum
Last updated: Sun Jun 18 10:16:42 2017		Last change: Wed Jun 14 14:59:20 2017 by root via cibadmin on Cluster-Tester-01

2 nodes and 3 resources configured

Online: [ Cluster-Tester-01 Cluster-tester-02 ]

Full list of resources:

 virtual_ip	(ocf::heartbeat:IPaddr2):	Started Cluster-tester-02
 cluster_vip	(ocf::heartbeat:IPaddr2):	Started Cluster-Tester-01
 zabbix_server	(systemd:zabbix-server):	Started Cluster-Tester-01

Daemon Status:
  corosync: active/enabled
  pacemaker: active/enabled
  pcsd: active/enabled

and also i need to know what i can do when the node 2 takes over node 1
i get the DB duplicates.

Thanks.