Inngest workflow is way too slow, is there any way to improve it?

[neoromantic]

So we have a job (this is an example)

You might notice that from 90s of execution, substantial part is delays of steps. Not even the run triggering delay, but every step execution delay.

This is crazy. Out of seconds of valuable work, we get 50+% of the time spent of (I reckon) inngest-based waiting.

Is it expected? Is it in any way normal? Can it be fixed?

[Dan6erbond]

I wonder if this is what's causing my perceived issue of steps "hanging"! I also noticed that sometimes I'll see step outputs of 1-2 steps, then nothing, and usually just give up. Are you running the dev server?