I'm using the haproxy template available here:
Zabbix server and agents just recently upgraded from 7.0.0alpha3 to 7.0.0alpha4. Problem was seen with alpha3 as well.
In the template, {$HAPROXY.BACK_RTIME.MAX.WARN} and {$HAPROXY.SERVER_RTIME.MAX.WARN} default to a value of "10s". I was seeing alerts saying the haproxy backend and server response time was over 10s for 5 minutes. Looking at a graph of the response times, it never goes above about 40ms.
It appears that the trigger math is not properly converting the value of "10s" in the macro to 10000 (to match the item unit) before it uses it in the trigger calculation.
Fixing it was easy ... I just added those two macros with a value of "100" and now it's quiet. Should the macro have been set to 10000 instead of 10s? Is the trigger math supposed to deal with different time units?
Zabbix server and agents just recently upgraded from 7.0.0alpha3 to 7.0.0alpha4. Problem was seen with alpha3 as well.
In the template, {$HAPROXY.BACK_RTIME.MAX.WARN} and {$HAPROXY.SERVER_RTIME.MAX.WARN} default to a value of "10s". I was seeing alerts saying the haproxy backend and server response time was over 10s for 5 minutes. Looking at a graph of the response times, it never goes above about 40ms.
It appears that the trigger math is not properly converting the value of "10s" in the macro to 10000 (to match the item unit) before it uses it in the trigger calculation.
Fixing it was easy ... I just added those two macros with a value of "100" and now it's quiet. Should the macro have been set to 10000 instead of 10s? Is the trigger math supposed to deal with different time units?