When using nested trigger function sum() and trendcount() in a trigger expression it appears that the time shift in my trigger expression is ignored. This is my trigger expression:

sum(trendcount(/test/test.test,3h:now/h-3h))>0

Steps to reproduce:

I've set the timezone on my system and the Zabbix Frontend to both be on UTC time and the current time is 19:55 UTC (Zulu). This should mean that the trend values used for determining if the above trigger is a PROBLEM or OKAY would be from the 1400, 1500, and 1600 hours. Without a timeshift parameter in the expression, the last 3 hours of trends would be 1700, 1800 and 1900.

[root@LAPTOP ~]# date
Sun Sep 19 19:55:21 UTC 2021

When I run the following script the trigger reports as okay in the web frontend, but this should actually yield a result of "1" from the expression above:

[root@LAPTOP ~]# for min in `seq 0 5`; do c=$(date --date="2021-09-19T*14*:${min}0:00" +%s); echo test test.test $c 95; done > file.out; zabbix_sender -i file.out -T -z 127.0.0.1
Response from "127.0.0.1:10051": "processed: 6; failed: 0; total: 6; seconds spent: 0.000054"

However when I change the timestamp parameter that I'm generating in the above script to be the 1700, 1800 or 1900 hour the trigger shows as a PROBLEM in the web frontend which seems to indicate that the timeshift is ignored in trigger expression

[root@LAPTOP ~]# for min in `seq 0 5`; do c=$(date --date="2021-09-19T*17*:${min}0:00" +%s); echo test test.test $c 95; done > file.out; zabbix_sender -i file.out -T -z 127.0.0.1
Response from "127.0.0.1:10051": "processed: 6; failed: 0; total: 6; seconds spent: 0.000054"

Is the timeshift being ignored or is there something else that is happening here?