Open selivan opened 6 years ago
Hello Pavel,
thanks for your article. I had a similar issue. Our zabbix randomly triggered hosts as unreachable for a few seconds without any logic reason. I tried to change unreachable/unavailable host settings and also timeout settings. Nothing helped resolve it.
So I changed Zabbix agent trigger Zabbix agent on {HOST.NAME} is unreachable from:
{Zabbix Agent:agent.ping.nodata(1m)}=1
to
{Zabbix Agent:agent.ping.last(0)}=0
and now everything works well.
edit: I tested this on v1.8.9
@pida42
Are you sure, that now it works as intended? Zabbix documentation says that agent.ping
value is 1 or nothing, you will never get 0.
And also if last value was collected a week ago, it is still considered the last(0), if I get it correctly.
@selivan Thanks, my mistake, it works only on older version (for me 1.8.9).
Finally did that using net.tcp.service
, only if monitored host is visible for Zabbix server on network.
Item:
Type: Simple check
Key: net.tcp.service[tcp,,10050]
Type of information: Numeric (unsigned)
Show value: Host availability
Trigger:
{Template App Zabbix Agent:net.tcp.service[tcp,,10050].last()}=0
Tested version 3.4.13.
@pida42 Have you tested it by blocking the host IP on server or by turning the host off?
I suppose, when host goes down, no new values for item net.tcp.service[..]
are sent to server. last()
returns the last seen value which is still 1, so the trigger is not switched.
EDIT: if the check is made from server host, not from agent itself, it should work.
@selivan
I did. I tried:
Yes, it's Simple check, so it runs from Zabbix server. It's enough for our solution.
Here is exported template: https://gist.github.com/pida42/126eeaee45ae303fe2ee549306a9208f
Thank you for your help...
This is the comment thread for Zabbix: unawailable items and hosts caused by wrong default timeout settings.