►
From YouTube: August 2021 OpenZFS Leadership Meeting
Description
At this month's meeting we discussed: OpenZFS Conference call for presentations
https://docs.google.com/document/d/1w2jv2XVYFmBVvG1EGf-9A5HBVsjAYoLIFZAnWHhV-BM/edit#
A
All
right,
hi
everyone,
let's
get
started
with
the
august
2021
open,
zfs
leadership
meeting,
don't
have
too
much
on
the
agenda
today,
so
we'll
have
lots
of
time
for
q,
a
or
other
topics.
A
The
main
thing
that
I
wanted
to
mention
was
that
the
open
zfs
conference
will
be
happening
again.
This
year,
it'll
be
online
again
november,
8th
and
9th.
A
The
registration
is
now
open,
it's
free
again
and
the
call
for
presentations
is
open.
So
if
you've
been
working
with
zfs
doing
some
development
discovering
using
zfs
and
discovering
how
it
works
or
anything
else,
you
want
to
share
your
experience.
We'd
love
to
hear
about
it,
send
me
an
email
by
september
3rd,
so
you
have
about
two
a
little
more
than
two
weeks
to
get
your
presentation,
your
proposals
in
not
not
too
long,
so
put
it
on
the
top
of
your
to-do
list.
A
My
email
address
is
on
the
main,
open,
zfs
webpage
and
all
the
dates
are
on
there.
Yeah.
So
we'd
love
to
see
you
all
attending
and
and
presenting
and
sharing
your
work
with
us
any
questions
about
that.
A
All
right:
well,
we
hope
to
have
a
few
presentations
about
the
zifes
on
object,
store
work
that
the
deflex
team
has
been
working
on.
Hopefully,
there'll
be
some
stuff
in
there.
That's
interesting,
even
if
you
have
no
need
of
zfs
on
object
store,
because
we've
kind
of
taken
some
taken.
Some
experiences
from
the
data
structures
and
algorithms
that
we
use
in
the
kernel
and
kind
of
redone,
a
bunch
of
them
for
other
aspects
of
the
object
store
and
the
cache
that's
used
with
it.
A
So,
hopefully,
we
can
kind
of
share
that
and
and
kind
of,
compare
and
contrast
and
maybe
learn
from
what
we've
done
there
and
take
some
of
that
back
to
the
core
zfs
as
well.
A
Cool
well,
I
think
that
was
the
only
thing
that
was
in
the
document.
What
other
things
would
folks
like
to
discuss
today.
B
I
guess
earlier
today
mark
maybe-
and
I
were
discussing
the
vw
property
stuff,
I've
rebased
it
and
we've
decided
it's
it's
past
the
work
in
progress
stage
now
so
we're
ready
for
people
to
look
at
it.
B
Yeah,
so
that's
the
stuff
I
presented
back
in,
I
guess
2018
and
2019
at
the
summits,
but
extended
with
mark's
stuff
to
do
basically
queuing
for
device
removal
and
we'll
be
able
to
set
the
no
allocating
flag
on
a
v
dev.
C
So
I
guess
one
question
is:
is
that
is
that
a
good
probable
possible
short
talk
for
the
upcoming
summit,
to
mention
that
work.
A
B
A
I
was,
I
thought
I.
A
B
A
We'll
have
landed
yeah,
hopefully
we'll
have
landed
by
by
the
conference.
B
B
Yet
that
was
extending
a
patch
paul
dodeck
did
like
a
decade
ago
or
so
in
freebsd,
but
making
it
work
for
more
properties
than
just
the
name
of
the
data
set
so
that
you
can
sort
by
the
you
can't
get
the
creation
date,
but
you
can
get
the
transaction
it
was
created
in
which
you
know
the
will
result
in
the
same
sort
order,
which
is
the
important
thing
because
only
being
able
to
list
the
snapshots
quickly
if
you
sort
them
by
the
name.
B
B
A
Well,
one
question
that
I
had
for
this
group
of
folks
was:
you
know
it's
been
a
few
months
since
we
added
the
additional
maintainers
for
the
project.
We
wanted
to
see
how
folks
feel
that's
going.
I
know
like
a
bunch
of
things
have
landed,
but
we
have
you
know
we
still
have
tons
of
outstanding
pr's.
So
I
guess
the
question
is
like:
are
pr
submitters
getting
the
feedback
that
they
need
to?
You
know,
get
their
work
integrated
or
you
know,
get
the
feedback
that
they
need.
B
B
Yeah,
like
I
think,
for
forced
export,
I
don't
really
have
a
roadmap
of
what
else
needs
to
be
done
to
get
it
to
be
ready
to
merge.
B
And
I
guess
the
same
thing
for
the
linux
namespace
support.
I
think
there's
one
open
question
about
what
happens
when
the
namespace
goes
away.
What
does
zfs
do
about
it?
And
I
don't
know
that
I
have
a
good
answer
to
that,
but
other
than
that,
I
think
that
one
could
use
some
more
feedback.
B
Although
I
know
you
know,
half
of
my
pull
requests
are
waiting
on
me
and
the
other
half
are
not,
and
I
can't
complain
but,
oh
sorry,.
A
I
mean,
I
think,
that
it's
reasonable
to
you
know,
bring
attention
to
the
things
that
need
attention
and
you
know
to
paying
reviewers
or
ping
the
this
I
need
the
maintainer
that's
been
assigned
to
it
to
to
get
that
feedback
yeah.
So
we
can.
I
I'm
looking
for
the
export
one
export.
Oh.
B
Yeah
but.
A
D
A
D
A
Sounds
like
maybe
the
we're
we're
doing
that
as
well.
The
four
of
us
working
together
are
doing
as
well
as
as
brian
did
on
his.
B
A
Ryan,
each
yeah
yeah.
Well,
you
probably
have
a
quarter
of
the
time
to
spend
on
it.
That
brian
did,
I
think,
he's
not
here
to
take
our
compliments,
but
hopefully
he'll
he'll
see
this
and
he'll
watch
it
later.
A
Yeah,
so
I
I
agree,
I
mean,
I
think
that
we
could
definitely
do
better,
I'm
glad
that
things,
some
things
at
least
are
getting
in,
but
I
I
agree
that
some
of
these
long
stand
things
have
been
open
for
a
while
and
ready
for
review
for
a
while.
We
could
do
a
better
job
at
getting
those
getting
feedback
on
those
because
they
they
tend
to
fall
through
the
cracks.
You
know
when
they're,
not
on
the
first
page
anymore,
they're
just
sitting
there
and
they're.
A
You
know
surrounded
by
a
bunch
of
other
pr's
that
maybe
are
not
ready
for
review
and
and
so
they
may
get
lost.
So
we
could
definitely
do
better
at
that,
and
I
think
we
need.
We
need
everyone's
help
right.
We
need
people
the
if
you
are
submitters
reminding
reviewers
and
maintainers,
and
we
need
maintainers
reminding.
A
You
know
reviewers
and
everything
else,
so
I
know
I
I've
been
a
little
distracted
the
past
month,
so
I
have
not
been
doing
a
lot
of
code,
reviews
or
bugging
of
people,
but
I
hope
to
get
back
to
that
soon.
A
Cool
and
I
that
the
force
x4
work
in
particular
I'm
interested
in
and
and
and
alexander
I
I
definitely
appreciate
all
the
performance
work
you've
been
doing
a
lot
of
it
is,
you
know,
as
you
know,
things
that
I've
touched
in
the
past
or
you
know,
worked
on
or
run
into
so
thanks
for
pushing
all
those
things
through
to
completion.
A
B
Well,
I
guess
kind
of
speaking
of
pushing
things
through
or
whatever
something
we
talked
about
last
month.
A
couple
of
these
I
thought
we've
talked
about
so
much.
I
thought
they
were
done
already,
but
we
talked
about
changing
some
of
the
defaults
like
enabling
lz4
compression
by
default
on
new
pools,
which
I
thought
we
agreed
on
like
two
years
ago
at
a
conference
and
then
it's
just
nobody's
ever
opened
a
pr
to
do
it.
B
I
guess,
but
we
talked
about
possibly
changing
the
default
for
compression
to
on
x
adders
to
sa,
where
that
makes
sense,
changing
the
default
block
size
for
z,
vols
from
8k
to
16k,
especially
in
and
possibly
if
it
makes
sense
for
the
default,
a
shift
to
be
12
instead
of
nine.
B
Since
the
cases
where
you
need
nine,
you
know
you
need
nine,
so
you
could
set
it,
but
maybe
the
default
should
be
higher,
and
if
it
is,
then
that
means
the
the
zval
default,
probably
really
does
need
to
change,
and
then
how
do
we
communicate
with
users
about
changing
all?
You
know,
four,
pretty
major
defaults.
A
Yeah,
those
are
good
questions
and
yeah
I
mean
for
some
of
those.
We
just
need
to
open
the
prs,
the
changing
the
default
block
size
landed.
I
think
today
did
the
eve
of
all
block
size
to
16k.
So
you
know,
maybe
we
can.
A
A
Yeah,
I
think
compression
defaulting
to
on
is
pretty
safe.
I
would
I
think
of
those
that
you
mentioned
the
default.
A
shift
feels
the
most
questionable
okay,
but
maybe
that's
just
because,
like
you
know
in
in
my
in
the
world
where
I
that
I'm
actually
operating
in
like
we
mostly
have
a
shift
nine
but
yeah,
I
mean
that
I
I
wouldn't
rule
it
out.
A
I
mean
it's,
you
know
it
is
supposed
to
be
automatically
detected.
I
would
say
like
if
there
was
a
way
to
say
if
we
can't
automatically
detect
it
then
default
it
to
4k
like
that,
would
make
sense.
But
but
saying
like
oh
the
disk
says
it
has
a
shift
9,
but
we're
just
going
to
use
a
shift.
12.
like
that
seems
potentially
not
great.
C
D
Yeah
on
on
on
freenus
and
trunas,
we
force
everything
to
4k
from
perspective
of
disk
replacement,
because
sometimes
some
customer
may
have
nine
but
then
has
to
migrate
to
12
just
by
disks.
From
that
perspective,
we
switch
it
to
12
quite
long
ago,
yeah
years
ago,
actually,
but
from
point
of
small
deployment
of
somebody's
laptop
or
from
point
of
space
efficiency.
Of
course,
a
shift.
Nine
is
much
better.
D
I
don't
also
in
case
of
rate
z
or
d
rate.
Difference
could
be
quite
significant,
but
I
don't
know
how
many
people
try
to
store
small
objects
on
wide
ranges
make
it.
A
All
right
other
topics
or
questions.
A
Now
it's
really
quick
meeting
everybody's
everybody's
taking
time
off
in
the
summer.
I
know
I
am
so
maybe
that's
why
we
don't
have
quite
as
much
stuff
today.
A
Hopefully,
next
meeting
we'll
have
more
news
on
the
open
gfs
conference
front.
Actually
is
that
right?
I
don't
know
if
we'll
have
speakers
selected
by
then
quite
by
then,
but
it'll
be
close.
A
A
Yeah
and
registration
is
open,
so
feel
free
to
do
to
go
through
that.
I
think
the
format
is
gonna
be
probably
the
same
as
last
year
in
terms
of
the
kind
of
technologies
that
we
use
so
like
you'll
need
to
register
to
get
the
zoom
link.
Otherwise
you
can
watch
it
on
youtube.
A
If,
if
folks
have
feedback
on
things
that
worked
or
didn't
work
that
they
didn't
give
last
year,
then
you
know
now
would
be
the
time
now.
Now
would
still
be
a
good
time
to
give
that,
if
you'd
like
to
see
something
happen
differently
with
in
terms
of
like
how
we
do
the
conference
around
the
conference,
the
technology
and
the
tools
and
stuff
seemed
like
people.
People
liked
the
like
zoom
breakout,
room,
socializing
aspect
of
it,
so
that
was
nice.
A
Yeah,
I
think
we
could
probably
do
a
better
job
of
organizing
the
hackathon
than
we
did
last
year.
I
think
that's
always
kind
of
a
challenge
to
get
new
folks
involved
in
that
when
they're
just
coming,
you
know
they're
just
seeing
seeing
a
few
people's
faces
on
zoo
on
zoom,
it's
tough
to
pull
them
in.
A
So
maybe,
if,
if
you
have
some
topic
that
you're
interested
in
that
you'd
like
to
kind
of
pull
in
new
folks
and
maybe
educate
people
on,
I
think
that
those
kinds
of
things
would
be
great
for
the
hackathon,
where
we
can
have.
You
know,
expert,
led
sessions,
or
maybe
not
even
expert
but
like
sessions
led
by
somebody
who
can
who
can
kind
of
direct
other
people
and
help
them
get
started
and
kind
of
know
what
to
do.
A
Give
them
a
good
good
like
hey,
come
to
this
session
I'll,
give
you
something
to
I'll
give
you
something
to
do
and
you'll
be
able
to.
You
know
hack
on
zfs
for
a
day.
A
A
All
right,
more
questions
or
topics.
A
All
right:
well,
then,
we'll
see
you
all
in
four
weeks.
I
think
that
will
be
the
earlier
time
on
september.
14Th
it'll
be
at
nine
o'clock,
pacific
and
yeah.
A
Think
about
what
you
would
like
to
talk
about
at
the
conference
and
send
me
an
email,
and
we,
I
think,
if
you,
if
you
don't
remember
you
know,
the
conferences
was
very
well
attended
last
year,
despite
being
online,
so
I
expect
that
we'll
you
know
hopefully
we'll
have
as
many
if
not
more
people,
and
so
it's
kind
of
up
to
this,
the
the
kinds
of
folks
who
attend
this
meeting
to
you
know,
share
their
work
and
and
provide
the
content.
A
A
So
I
look
forward
to
emails
and
I'll
see
you
again
in
four
weeks,
see
you
everyone.
Thank
you.