Rob Norris
Rob Norris
Block and record sizes inside OpenZFS are nearly always stored and manipulated as bit shifts, not absolute sizes, and are stored on disk as well. That alone would make this...
> I agree with @robn's [comment in the original PR](https://github.com/openzfs/zfs/pull/16591#issuecomment-2398001976) Since I was invoked, I will say that I am not and never was against this change _as such_. What...
> Before I even looked into the current patch, I think instead of calling a I/O queue control, we should call it a I/O scheduler control. +1. I didn't want...
Dang, what a fascinating bug! Thanks for the report! If it ever becomes an issue (like, some maintstream distro doesn't end up taking the patch) we can work around it...
Ahh, clang! Confirmed, working on it. Cheers.
If you'd like to test, #17954 should get things going again. Thanks for the report!
Last push rebasing to master to get fresh test results.
I am uncomfortable with this change. I agree that the `vq_lock` runs white-hot when the disks underneath are very very fast. I've spent many months trying to reorganise this lock,...
@tonyhutter thanks for trying to test those cases; I'm mostly going from memory and gut feels and its good to have things confirmed, or not! From memory the deadman path...
(incidentally, I love how many subtle and weird and things properly fast IO is throwing up. It's nice to be solving hard problems!)