Hi,
we're monitoring our servers from another data center via Internet. When the provider's internet link, where the Zabbix Server is located, has a problem (i.e. packet loss or even complete outage) we still get some false alarms after tweaking our config.
In the events list we see the link seems to have a problem because many servers turn red until finally www.google.ch is checked and turns red, too. The link is down for more than 5min and a notification for all the hosts which turned red before www.google.ch was checked is sent out.
We don't know if this behavior is a bug or if our configurations still lacks some tuning. We think that before the notification is sent the status of the depending trigger should be checked again to avoid such false notifications.
These forum articles go in the same direction:
Any ideas/help/comments are greatly appreciated.
Best regards,
Andy
we're monitoring our servers from another data center via Internet. When the provider's internet link, where the Zabbix Server is located, has a problem (i.e. packet loss or even complete outage) we still get some false alarms after tweaking our config.
- We use Zabbix 1.8.2
- The action is set up to send an SMS only if the host/service is down for at least 5min.
- All hosts are dependent on the reachability of www.google.ch which will be checked quite often by an icmpping simple check.
- Trigger for www.google.ch: {www.google.ch:icmpping.avg(300)}#1)
So Google has to be fully reachable for 5min before any depending trigger may turn red - Some servers use the Zabbix Agent. For these hosts we never received false alarms.
- Other servers are checked using simple checks (basically the standalone template). These are the "problem hosts".
In the events list we see the link seems to have a problem because many servers turn red until finally www.google.ch is checked and turns red, too. The link is down for more than 5min and a notification for all the hosts which turned red before www.google.ch was checked is sent out.
We don't know if this behavior is a bug or if our configurations still lacks some tuning. We think that before the notification is sent the status of the depending trigger should be checked again to avoid such false notifications.
These forum articles go in the same direction:
Any ideas/help/comments are greatly appreciated.
Best regards,
Andy