Tasks checkpoint at every 10% when a trickle message is sent is sent back to the server. Assuming a task takes 12 hours on your machine back of envelope maths 12×60 = 720 720÷10 = 72 this means your task will checkpoint roughly every 72 minutes. The deadline for tasks is 4 days I hope this helps

Edit = added task deadline

Thanks for the info. I guess this research is not for me. I'm a kinda put the pc to sleep but always shutdown like every 2 or 3 days. I can't really remember if it was 6 hours or 12 hours. But I guess I'm better off with shorter research with shorter checkpoint.

I also learns new stuff from this info.

I thought check pointing are like save state, which they write research memory content to disk where they can recover from, like VirtualBox save state.

Anyway, thanks for the info!

FAH2 checkpoints frequently, about every 10 min on my boxes. runtime between 16 and 20 hours. trickle sent every 10%, about 2 hours for me. I have stored the trickle msg for about 10 hrs and sent a bunch at once, only problem found is that 90% must be sent and processed by server before final upload or job fails validation