trying to set up an alarm for pods stuck in pending status for more than 5 minutes, but I generally get alerted within 2 minutes instead (even when alert condition is set to "all the times" during the last 5 minutes. Any ideas?

It might be possible the for a window of 5 minutes, for the first 3 minutes there is no data; just when the pod came online in the last 2 minutes, it is in the pending phase and you got alerted?

yes, normally that's the case

I think a new feature needs to be added where you can configure to alert only if the data exists for full window.