anybody interested in picking up hreinecke/sg3_utils #58 ?

jlevon: it looks like (from a quick glance) that the 'upstream' (I guess, i never quite understood the relationship between the two sg3 utils repos on GH) has that fix

(is it a fork? is one 'abandoned', etc)

I think at the time, that didn't exist. thanks

yeah, i have no idea what the relationship is there other than the one he references seems to have the change..

i've been able to build them and use them w/o any problem

it's been helpful for all of my SATL changes (I still have a few more I need to get up for review)

also helpful when trying to figure out how broken VMware's scsi emulation is

(sadly their NVMe emulation isn't any better)

I do also have a change I could upstream (at least I don't think i've upstreamed it yet) where drives that don't advertise support for UNMAP, but do advertise some other bits, it will issue WRITE SAME (with zeros) CDBs for DKIOCFREE ioctls instead of issuing UNMAP CDBs

which on VMware at least, will shrink thin-provisioned vmdks

[illumos-gate] 5915 msgpullup(9f) behaves as if len is always -1 -- Kyle Simpson <kyle⊙oc>

pardon my off-topic rant: I hate AFCI circuit breakers. (the circuit my illumos dev machine is plugged into was taken down by a false arc fault trip when a vacuum cleaner stopped).

:(

hopefully nothing lost...

jclulow: About c39t5000CCA04D1A6BA9d0 = it is another system with another case with changing c39->c38

vetal: Right, I think we want to get all of the details that we can about that system so that we can figure out what's going on

jclulow: However both path with c38 and c39 still exist. I think you are right and it is multipathing cases

jclulow: gist.github.com/gusev-vitaliy/7c29ae481964e20dc933d7d4f45888cb

jclulow: However I am confused why diskinfo reports only one path.

diskinfo does some deduplication based on media identity

It gets a bit confused if two disks are apparently identical, I think

vetal: Can you grab all the other commands I listed on that system as wel l

tsoome: just my time..

[illumos-gate] 16854 CPUID_PASS_IDENT is sufficient for cpuid_getsig -- Luqman Aden <luqman⊙oc>

jclulow: fmtopo-Vp.txt file.io/Ita7LGuClvkh

vetal: It says it's been deleted

jclulow: Interesting. Sorry for that. file.io/jWwj65leCTV3

jclulow: "File will be deleted after download". It seems it is only for one download.

vetal: I've created a bug now: illumos.org/issues/16937

→ BUG 16937: multipathing not engaging on SAS JBOD? (New)

It includes a list of things we should gather as part of investigating

if you can whack the files in there that should be easier hopefully

jclulow: Will provide required files tomorrow.

Ok, thanks!

jclulow: Thank you too!

You're welcome! Hopefully something will jump out in the data and it's an easy fix