Hello,
I'm using zabbix 3.0.8 in virtualbox, combined with some active zabbix-agents running on windows.
These agents are monitoring free diskspace of other remote windows servers through performance counters. (using template & item/trigger prototypes)
In a discovery rule (5m interval) there is an item prototype with:
And a trigger-prototype with:
The idea of the trigger is to detect when free diskspace suddenly starts dropping faster than expected by comparing the currently free diskspace with that of a week ago.
The problem is that the trigger sometimes comes in to Problem state, even when the counter returns a flat 35.38GB free space over the previous weeks.
I also noticed it stays in Problem state 5/6 minutes and then reverts back to normal. I have no idea if this is a coincidence, or if the discovery interval can have an impact on the fireing trigger. The discovery of the disk is just returning a json-file locally available to the agent, so discovery should work every time.
Can this be due to missing data for a couple of minutes/hours? I included the min(1d,7d) to avoid having 'empty' reference data.
Or any other ideas?
I'm using zabbix 3.0.8 in virtualbox, combined with some active zabbix-agents running on windows.
These agents are monitoring free diskspace of other remote windows servers through performance counters. (using template & item/trigger prototypes)
In a discovery rule (5m interval) there is an item prototype with:
Code:
key: perf_counter[\\{HOST.NAME}\LogicalDisk({#DSK_NAME})\Free Megabytes]
type: Numeric(float)
Units: B
Multiplier: 1048576 (because the performance counter returns MB)
intervals: Update: 60s History: 90d Trend: 365d
Code:
{Template Disk-Discovery:perf_counter[\\{HOST.NAME}\LogicalDisk({#DSK_NAME})\Free Megabytes].min(1d,7d)}
-{Template Disk-Discovery:perf_counter[\\{HOST.NAME}\LogicalDisk({#DSK_NAME})\Free Megabytes].last()} > 500M
The problem is that the trigger sometimes comes in to Problem state, even when the counter returns a flat 35.38GB free space over the previous weeks.
I also noticed it stays in Problem state 5/6 minutes and then reverts back to normal. I have no idea if this is a coincidence, or if the discovery interval can have an impact on the fireing trigger. The discovery of the disk is just returning a json-file locally available to the agent, so discovery should work every time.
Can this be due to missing data for a couple of minutes/hours? I included the min(1d,7d) to avoid having 'empty' reference data.
Or any other ideas?