Convert distributed monitoring into centralized monitoring

[fatiha]

Hi,

I have to convert distributed monitoring (a master and 3 slaves) into centralized monitoring (only a master). How can I do that ?

I changed parameters in conf file like "Server=" and it works. But I don't know how to stop synchronization between the 4 databases ??

Thanks very much,
Fatiha

[ataylo13]

I am not sure I understand the syncronization issues, but if you configure the agents to send data to the master and shut off the slave zabbix_server processes there will be nothing to sync.

[fatiha]

Ok, thanks for your answer.

[fatiha]

Do not stop zabbix_server in a node because the master doesn't like it !!!

I proceeded to the migration of the servers of a node towards the master then I deleted hosts on the node (including the updating of zabbix_server in zabbix_agentd.conf file)

At the moment when I stopped one of the nodes, there was a total desynchronization of the information and the master stopped brutally.

It was impossible to restart the master, who was instable. So I restored the master's database and the nodes' databases and I lost all actions, maps and it services

So what can I do now ???

Fatiha
ps : I am using zabbix 1.4.4

[fatiha]

Can someone help me ?

How can I stop nodes without stopping master ?

Look at zabbix_server.log :

27527:20090505:193619 One child process died. Exiting ...
27527:20090505:193621 ZABBIX Server stopped

But none of the zabbix_server's nodes stopped !

I am trying to restore old backups but the problem was not solved.

Thanks for your help,
Fatiha

[ataylo13]

I am not sure I understand your issue. I will preface this with I am using 1.6.4 and not 1.4.4.

1. From the remote node i have exported the hosts to xml.
2. Shutdown the zabbix_server on the remote node
3. go into administration and remove remote node from the master's node list
4. import xml of remote nodes hosts to master node
5. configure as you normally do
6. repeat on other nodes

Doing it this way you will lose all historical data for the remote hosts.

HTH

[fatiha]

Thank you, it works fine now. I haven't deleted nodes in the administration's interface, that's why it didn't work.

Thank you very much !

Fatiha

[ataylo13]

No problem.

[fatiha]

Hi,

Maybe it is a zabbix 1.4.4 problem.

I made those tests :
- remove nodes like you said (6 steps) => it works fine
- stop zabbix_server
- backup of master's database (with mysqldump)
- restore master's database (with mysql)
- start zabbix_server => KO

12516:20090506:194003 One child process died. Exiting ...
12516:20090506:194005 ZABBIX Server stopped

These steps are important because I want to install zabbix 1.6, but without a good backup, I don't know how can I do.

Can you help me please ?

Thanks,
Fatiha

[ataylo13]

Do you have the Master server in debug mode? I would try to find out what child process is dying causing zabbix to crash. I have not had 1.6.x crash when a child node was unavailable, it will be listening but not receiving anything.

[fatiha]

Hi,

I put zabbix_server in debug mode, so I have those logs :

20338:20090507:105012 In update_trigger_value(triggerid:100100000015629,old :2,new:2,1241631483,ZABBIX was down.)
20338:20090507:105012 End update_trigger_value()
20338:20090507:105012 Query [select min(i.nextcheck+i.delay) from hosts h,items i,triggers t,functions f where f.triggerid=t.triggerid and f.itemid=i.itemid and h.hostid=i.hostid and i.nextcheck<>0 and t.triggerid=100100000015631 and i.type not in (2)]
20338:20090507:105012 In update_trigger_value(triggerid:100100000015631,old :2,new:2,1241631485,ZABBIX was down.)
20338:20090507:105012 End update_trigger_value()

I have a master and three nodes.

-----------------------------------------
Today I have changed zabbix_server.conf, instead od Node=1, I put Node=0.

And in the web interface, I have created a new node "master" with id=0 and I have migrated all hosts from the old master (id=1) to the new one.

-------------------------------

There's no more actions printed when I created the new node with id=0.

It's strange because the views "latest data" and "triggers" are fine. But no actions, no mails sent ...

Maybe I will install zabbix 1.6 from scratch because I lost an entire week to try this conversion (distributed monitoring to centralized monitoring).

If someone has done that conversion with zabbix 1.4.4, please tell me how you did.
__________________

Thanks,
Fatiha