FYI: I have migrated the bug tracker to new hosting. Let me know if you see anything weird with Redmine.

it is snowing:)

redmine is snowing?

nooo:)

we have first snow:)

hanging out for the first big summer thunderstorm here

australia?

is there supposed to be a storm today?

I would like even a thimble full of water to be spritzed over SF at this point

:D

dlg: not that I know of, though they said there was maybe going to be one yesterday, then it blew away

tsoome (IRC): yeah, .au

morning

moin

Morning

tsoome: we have snow with rain

yea, it is changing here too, it is october afterall:)

my laptop is puzzling me... somewhere it drains space from my pool, and I can't put a finger on it (likely regular znapzend snapshots play a role, but not proven yet - and on the dev workstation I don't think I have that much data traffic to make those differences)... last time a few weeks ago when it ran out of space completely, I rewrote (zfs send|recv) a number of datasets with likely dedup sweet spots, hit about 3x savings on those

(by zfs -DDD summary), and freed up 60Gb, now back to about 15Gb usable (per df) and about 40Gb free in pool (per zfs list)

with all that, a `zfs list -tall -d1 -s used -r nvpool` puzzles me yet more: "USED" of the pool equals its whole size per `zpool list`, but sum "USED" of the datasets is 2/3 of that

it is a single-device pool (remainder of a mirror) so no raidz math involved

details here FWIW : pastebin.com/hAGPfkva

hm. what does zpool status say?

well, it is degraded and faulted since it is a broken mirror, but otherwise "no known data errors"

it actually never succeeded to be a mirror... being a VM under a Windows 10 host, it intended but couldn't use a partition on system disk as part of the pool setup

the devices can be allocated but not written (win7 could)

fwiw updated znapzend to current master, will see how it goes (to clean up older snaps), if that helps

will take some hours

ideas welcome in the meanwhile :)

probably then will also update OI, maybe some accounting or deferred cleanup or whatever is borked in recent months' builds?..

my current root is from a couple of months ago: nvpool/ROOT/hipster_2020.04-20200809T192157Z

Silly question, I can't find a filed issue, but has anyone complained about the inability to ifconfig or ipadm a x.y.z.N/31 IPv4 address (or a LLLL:MMMM::NNNN:OOOO/127 IPv6 one)?

There IS a way to configure an ethernet NIC to do it, but it's completely non-obvious:

ifconfig e1000g0 x.y.z.N destination x.y.z.(((N & 0xFE) == N) ? N + 1 : N - 1) netmask 255.255.255.255

I've never heard any complaints about that. For P2P links I've always used at least a /30 anyway

but allowing /31 seems like a good idea

danmcd: It'd be good to have the point-to-point support in ipadm, yes

It's there in ipadm too, just not in the /31 format.

Ahh

-a local=x.y.z.N -a remote=....

Ah, so -a XXXX/31 would be sugar, of sorts, for not using two totally different addresses as the local and remote

Point-to-point configurations are arbitrary yes, and a /31 is a subset of the point-to-point configurations.

(Pardon latency, was helping with dinner prep, now off to wait for cook-to-finish and eating.)

Makes sense

I'm filing a bug.

illumos#13232