Re: stable/15-aX file/IO issue after compiling 500 ports

i've seen no such file errors on zfs before

i don't know if it was git i just ran. others told git runs in  bg. didn't check. tho what does git do in *obj* prefix

it made no sense, i ran full clean obj buildworld, was greeted with no file/dir errors. puzzled, i ran again and it worked

this is even on 13.*

zfs also seems to get slow, and fills up wired memory with non-arc data, that could result in kills

no hw errors reported

very funnily, despite this being piss poor hw, if i just warm reboot it, it's completely fine immediately

i haven't seen any kind of actual data corruption, just file went somewhere and then it come back. i didn't check what or how but building is often ton of fs ops, delete, create, append, overwrite

i'm not fs expert, i'm only wtf'd by effects

yet, if i tried to push zfs over edge deliberately, by doing definitely things that surpass the capability of hw, there was nothing slow. it's even able to run find over million files in 30s if cached

i have never looked what's inside zfs test suite, possibly a thing that would cause zfs to fail if it has a issue

on ufs i ever once recall ending up with 0b file in power failure. this was my single ufs issue and that's not even software problem

here, i don't know. all i have to praise is how zfs has gotten a lot better in 12 years. i had 40m/s speeds, etc. i was like my hw can't possibly be that bad. and now i get 200m/s on worse hw

file not found, tho?