richlowe: Yeah I think it kind of has to do that unfortunately

I had hurry yesterday upgrade to r151046.. but I had issues with pkgsrc branded zones. Upgrade would not pass without detaching those zones and upon booting into upgraded global zone the pkgsrc won't attach (no errors though)

I feel rather confident though that I've managed to solve this maneuver before, but I can't remember the steps for success

duh, zoneadm -z <NGZ> -u did it (not that visible in the man page) :-)

is there a reason other than history that some things (idnkit, nspr, nss) still have header packages, rather than the headers in the main package like normal?

i think it's history

it'd be good to clear up

the variants on the headers should let people turn them on/off as they wish

and it would deal with things where we tried (and failed) with pkg to have a given fmri either be a leaf, or not.

so pkg:/library/idnkit and pkg:/library/idnkit/header-idnkit for eg is weird

(it's possible we failed at that to such a degree nobody knows we tried...)

yeah we tag headers etc already with the devel facet. so that should be good enough

is there an equivalent of `journalctl -b -1`? my home server just seemingly randomly restarted and i want to know why or what happened

not a direct one that I know, but we should log when we come back up to say what happened if we have crashed.

i was hoping there would be something in dmesg but im not sure what i was expecting to find there

hpqoeu: if setup there should be a stacktrace and a kernel dump Checkout /var/adm/messages

thank you, i think i found the reason

though i am surprised one of the disks being bad enough to hang i/o on a zpool would bring down the whole system

It can happen when the disks triggers the data integrity deadman

that makes sense, the log says "zfs:vdev_deadman"

Yeah thats a very missbehaving disk

yeah, i was already aware it's failing, was just hoping that zfs would fault it before it got really bad

but might be time to replace it either way

Hrd to say what will happen first. we usually replace once we know it's faulty

yeah i already got checksum errors on it, but the array doesn't hold important data so i was just putting it off

thank you for the help though, appreciated

yeah, ideally the disk faults, but in circumstances where it can't, the deadman panicks to reboot and get things back in service (ideally, the disk wouldn't import, when it reboots, but disks are disks...)