[illumos-gate] 18377 hma conflates CPU ID with sequence ID -- Andy Fiddaman <illumos⊙fn>

well that was annoying...

maybe because the window was open, but redmine seems to have eating my bug

guess i'll try again...

there we go

people might be interested in this one...

huh

fenix illumos#18384 perhaps ?

BUG 18384: ZFS autotrim can corrupt pool (New)

↳ illumos.org/issues/18384

yeah

(just turn it off if you have it on).... you probably won't hit it though unless your pool is really busy

since you need it busy enough that the TRIM I/Os end up sitting in the scheduler for at least a little bit

(and a full enough pool that zfs will want to allocate from the metaslab it just re-enabled moments ago vs. one of the others in the pool that weren't disabled)

i'll try to get the fix up in a bit

though I'm not sure about reproducing this -- it was found while looking for another issue

but also I think it's one of those that'd also be correct by inspection based on the earlier fix

wiedi_: hwgrok compiled, but for a bunch of my machines the drive bay info doesn't parse right

its lame.

sad :/

since it should all come from the same place, does it also show up wrong in diskinfo?

probably.. IIRC diskinfo ultimately uses the same topo enumerators

yeah its some issue with my /dev/es/ses devices

i compiled and installed sg3_utils which does get the correct info

On that note, this is a weird error: cannot set property for 'rpool': property 'autotrim'(31) not defined

for got to pfexec the zpool command but certainly not the error I was expecting

maybe pool needs updating?