Hmm -- this is baffling. Trigger Holdoff does indeed not work for me either, reproducibly, under pretty simple conditions. Please see the attached screenshot.
Test signal has 100 µs bursts of a 100 kHz sine-wave, at 1 ms intervals, coming from an SDG2042. I have set the holdoff time to 200 µs -- longer than the burst duration, shorter than the interval, which should get stable edge triggering at the first edge of each burst. Shown in the screenshot with 1 second persistence time to make some of the false triggers visible.
The scope does seem to respect the holdoff time as it should: When I measure the waveform update rate via Trigger-Out and an external counter, it drops as I increase the holdoff time. At very long holdoff times (say 200 ms), the frame rate drops to where you would expect it to be (5 Hz). But I can't get stable triggering. What am I overlooking?!