Zabbix poller processes more than 75% busy

[fanf074]

Hi all,

I configure ZABBIX to monitor 40 Hosts (Windows servers) and i have this alert in my dashboard : Zabbix poller processes more than 75% busy

Does anyone have a solution for this problem ?

ZABBIX run with the ZABBIX 2.0 VIRTUAL APPLIANCE (VMWARE).

Thanks for your answers.

[jix]

!

Try to increase your start poller in zabbix server configuration

[fanf074]

Hi all,

Even if i put the zabbix start pollers @ the max (1000), i have always the poller busy @ 100% ! It's change nothing for me.

Ave you another idea ?

[cheezus]

You can forget trying to get help on these forums. Zabbix is cool but performance is awful and building a server from scratch with it is hard as heck. We've never been able to to do it 100% correctly with all things like snmp traps, etc., working.

Posting on here is like pissing in the wind. They want you to buy their support plans so don't answer. Wish I could help but I'm in the same boat which lead me here. Unfortunately we're going to have to ditch zabbix because it's it's performance is severely lacking.

Don't get me wrong, we used to love zabbix. But these forums are nothing but dead end threads of issue, after issue. The support plans have become outrageously priced like zenoss did. It's sad, really, as we were once vocal supporters of zabbix

[jix]

!

im agree with cheezus
forget trying to get help on these forums
but the way

tell me about your server status

number of item?, active triger? , NVPS?

[fanf074]

Hi I'am a french Guy,So the dashboard will be in french.

Here is the Dashboard :

Serveur Zabbix en fonction Oui localhost:10051
Nombre d'hôtes (surveillés/non surveillés/modèles) 65 41 / 0 / 24
Nombre d'éléments (surveillés/non surveillés/non supportés) 2587 2405 / 0 / 182
Nombre de déclencheurs (activé/désactivé)[problème/inconnu/ok] 1234 1232 / 2 [2 / 0 / 1230]
Nombre d'utilisateurs 3 1
Performance serveur requise, nouvelles valeurs par seconde 39.2 -

Have you any idea ?

[jix]

!

change your start poller to 50, and check your poller status...

[ahowell]

You can forget trying to get help on these forums. Zabbix is cool but performance is awful and building a server from scratch with it is hard as heck. We've never been able to to do it 100% correctly with all things like snmp traps, etc., working.

Posting on here is like pissing in the wind. They want you to buy their support plans so don't answer. Wish I could help but I'm in the same boat which lead me here. Unfortunately we're going to have to ditch zabbix because it's it's performance is severely lacking.

Don't get me wrong, we used to love zabbix. But these forums are nothing but dead end threads of issue, after issue. The support plans have become outrageously priced like zenoss did. It's sad, really, as we were once vocal supporters of zabbix

I'm not sure what the performance problems your talking about are. We doing 680 vps with no issue what so ever on VMs with small amounts of memory and CPU, and yes with have SNMP traps working.

I've found the support from the community on the forums to be quite good, and the official guys to respond quite quickly even without a support contract if you open a bug with sufficient detail.

[bga83]

change your start poller to 50, and check your poller status...

Is there anything interesting in logs? It sometimes useful to increase Timeout parameter in zabbix_server.conf

[fanf074]

I have always this problem . I will try to increase the timeout in the zabbix server.conf.

[jhrasko]

maximum db connection reached

I have always this problem . I will try to increase the timeout in the zabbix server.conf.

If you are using MySQL database try to increase max_connections parameter. When you start many pollers they create many DB connections and standard MySQL limit is 150. For example set it to 500.

[cheezus]

BTW - We upgraded to 2.0 and changed all the agents to push instead of the server to poll and it's working solid so far.

[nicolasg]

increasing the number of Zabbix pollers from 5 to 100 seems that solved the issue in my case.

[BDiE8VNy]

[...]i have this alert in my dashboard : Zabbix poller processes more than 75% busy[...]

Just got that Trigger fired too after upgrading from 2.0.2 to 2.0.3 via RPM.
I used to source a file (/etc/sysconfig/zabbix) within the zabbix-server init script which sets some http_proxy variables. The init script has been replaced by the update and my web service checks run against the (fire)wall.
Adding the missing source command solved that issue (at least in my case).

[marmeks]

I increased:
Zabbix_server.conf -> StartPoller from 5 to 50.
MySQL conf Max_connections from 200 to 400.

Restarted MySQL and Zabbix service.

Trigger on "Poller busy 75%" is gone.
So are gone many false positives on unreachable hosts.

And Zabbix queue (Administration -> Queue) looks much more better.

Thanks for advices!