►
From YouTube: 2022-04-05 Rook Community Meeting
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
A
Hello,
everyone
welcome
to
the
work
community
meeting
for
april
5th
2022.
we'll
go
ahead
and
get
started
talking
about
the
1.9
release.
Let
me
see
if
I
can
share
my
screen
with
the
agenda
all
right.
A
There
we
go.
I
guess
the
first
first
few
comments
on
the
1.8
release,
so
we
did
have
the
1.8
release.
Last
thursday,
the
31st
and
1.8.9,
I'm
thinking
we
can
probably
move
to
a
three
or
four
week
cadence.
So
the
last
one
was
three
weeks
apart.
A
It
has
been
four
or
five
months
since
the
1.8
release.
I
think
we're
getting
pretty
stable
there,
not
too
many
backboards,
so
maybe
april
21st
we'll
shoot
for
that
for
the
next
one.
Unless
there's
anything
that
is
more
urgent
that
comes
up,
I
don't
think
there
was
anything
in
the
board
that
was
interesting
to
look
at.
Let
me
just
double
check.
A
Yeah
these
two
issues
don't
even
seem
like
1.8.
We
can
probably
just
move
them
out
to
1.9
I'll.
Take
another
look
at
those
but
yeah,
nothing
too
blocking
there.
Unless
anybody
else
wanted
to
bring
something
up.
A
A
We
currently
do
have
the
the
dot
17.1
release
out,
which
we
are
testing
against
in
our
ci,
but
just
to
make
sure
we're
good.
With
the
final
release.
I
think
it'd
be
good
to
wait
a
few
more
days
for
the
quincy
release
to
be
out
before
we
release.
A
So
1.9
our
beta
release
was
the
branch
was
created
and
all
that
and
the
release
is
out.
I
haven't
heard
any
feedback
on
it,
yet
it
is
out
there
for
testing
and
then,
if
we,
so
we
were
targeting
this
thursday
in
two
days
and
I'm
thinking
or
the
proposal
is
to
move
it
out
to
next
tuesday,
the
12th
just
to
give
quincy
a
little
bit
more
time
and
us
to
respond
to
it.
B
A
A
B
A
A
There's
csi
encryption
support
on
a
per
pv
encryption.
I
think
that's
pretty
close
and
it's
in
final
review.
I've
needed
it
to
take.
You
think,
take
another
look
at
that
pr
to
see
if
it's
actually
raiding,
but
that's
yeah.
That
may
perfectly
be
fine
to
come
into
the
patch
release,
not
technically.
A
And
then
the
admission
controller,
starting
up
by
default
without
any
script,
this
will
be
really
nice
to
have
because
we
can
finally
get
the
emission
controller
as
long
as
the
cert
manager
is
installed.
If
the
cert
manager
isn't
there
we'll
just
say
it's
optional,
we
won't
configure
it,
but
this
this
will
be
good
to
finally
have
an
admission
controller,
that's
easier
to
start.
Instead
of
with
a
separate
that's
in
review
too.
A
A
B
B
Probably
yeah,
I
guess
we'll
see
how
the
work
goes
on
that,
whether
we
think
it's
ready
for
ga-
or
perhaps
you
know
just
saying
at
least
the
api
kind
of
components
of
this
are-
are
stable
in
1.10
there.
You
know,
there's
time
until
then,
yep.
B
Yeah,
I
I
think
it's
more
just
things
around
just
kind
of
how
things
will
work
in
general
right.
A
B
B
A
B
B
B
Yeah
and
then
I
think
there
may
be
some
kind
of
subtle
changes
to
the
documentation
we
have
now
because
of
this
or
there
potentially
could
be.
So.
I
just
want
to
make
sure
it's
like
nice
and
polished
across
the
docks.
More
broadly.
A
C
A
B
Yeah,
I
I
saw
a
task
in
there
about
multi-site
realm,
so
I've
been
chatting
with
alimuridia
more
the
past
few
weeks
about
multi-site.
I
think
he's
going
to
be
he's
gone
through.
I
think
he's
had
four
multi-site
issues
and
triaged
them
from
from,
like
his
view
and
he's
going
to
be
kind
of
working
on
those
and
also
updating
some
docs.
B
Multi-Site,
so
I
guess
that's
just
to
say,
like
I
think,
multi-site
has
not
really
had
a
lot
of
things
changing
in
the
past
like
year
or
so,
but
I
think
we
can
expect
some
more
updates
coming
in
the
next.
A
Sounds
good,
okay,
that's
all
for
the
1.9
board.
Then
I
guess
on
the
agenda
items.
I
just
have
a
carryover
from
last
time
that
I
have
something
to
follow
up
on
where,
in
quincy
the
the
built-in
pool
device
health
metrics
was
renamed
to
dot
manager
and
with
the
upgrade
guide
we
just
want
to
make
sure
that's
tested
there
to
make
sure
we
don't
need
anything
else,
special
or
if
seth
just
takes
care
of
that
itself.
During
the
upgrade.
B
Yeah,
I
guess
yeah,
because
I've
like
started
thinking
about
the
upgrade
guide
a
little
bit.
B
So
I
don't
think
rook
does
any
handling
to
internally
change
its
like
idea
of
device
health
metrics,
like
being
the
name
of
a
pool
like
automatically
switching
to
mgr.
I
hope
we
don't
need
something
like
that.
That
would
be.
I
don't
yeah,
just
the
special
stuff
upgrade
conditions
like
that
are
annoying.
B
C
A
D
A
1.8
release:
we
have
this
feature
which
allows
you
to
actually
like
set
the
replicas
and
other
config
on
that
pool,
specifically
the
built-in
pools
and
and
if
they're,
creating
that
cr
that
so
during
the
upgrade
seth
will
rename
the
pool,
but
then
rook
might
just
create
the
old
pool
again.
I
think
that
may
be
the
only
side
effect
unless
you
rename
the
cr.
B
C
At
least
it's
not
too
big,
normally
deadpool.
B
A
A
A
C
C
Well,
well,
I've
basically
switched
jobs
yeah!
Well,
I'm
now
a
founding
engineer
of
co
technologies
and
well,
I
can't
necessarily
say
more
than
that:
we'll
do
something
with
rook
and
well.
I
think
I
should
keep
it
at
it
for
now
and
finally
get
the
website
up
and
running
with
the
actual
content.
C
A
A
D
Yeah,
I
guess
I
guess
I
might
try
to
add
some
proposed
to
add
some
discussion
topics.
I
still
have
some
questions
regarding
golang
and
used
language
features
for
rook
any
intersections
with
the
library
called
json
dash.
Iterator
is
one
question.
Another
question
is
about
reflection,
as
actually
is
the
reflect
package
and
probably-
and
I
got
one
one-
recurrent-
probably
two-
probably
a
recurrent
topic
related
to
risk
five.
D
I
got
some
access
got
some
access
for
the
purpose
of
the
go
go
levien
project
from
chinese
chinese
people.
There
is
a
option
to
remove
the
operates.
The
some
boards
I
didn't
check
yet,
but
I
just
only
had
some
they
provided
an
account
so,
but
that
I
didn't
check
yet
and
I
and
it
wasn't
about
rook
generally,
but
there
is
some
possibility
to
to
use
to
use
some
and
a
few
risk
fire
boards
to
run
some
concrete
tests
out
there.
D
Plus
there
is
some
support
for
kubernetes
and
docker
on
rs5,
so
I
thought
it
might
be
sort
of
like
an
option
to
check
out
about
potential
porting
of
maybe
if
there
would
be
any
interest
to
port
rook,
to
risk
5
and
to
see
which
technological
issues
might
be
there.
D
So
the
end
the
hardware
people
can
propose
maybe
some
some
initiatives,
but
they
have
to
understand
what
is
supported
generally
so
so
I
have
here
so
so.
This
is
what
I
wanted
to
like,
maybe
to
propose
to
discuss
so
like
like
that
yeah.
A
For
the
risk
five
support,
what
would
it
take
to
get
machines
available
to
add
to
the
ci,
so
we
can
run
them
with
github
actions?
I
think
you
know
anytime.
We
have
a
new
platform,
that's
the
biggest
risk
is
just
being
able
to
if
we
can
add
it
to
the
ci.
We
know
it's
tested.
We
know
it's
supported.
D
Yeah,
generally
speaking,
they
ask
about
port
distro,
so
you
can
assume
any
officially
supported
linux
distribution,
it
might
be,
ubuntu
might
be
fedora,
it
might
be.
You
can
check
out
about
opensuse,
I
guess.
D
As
for
other
links
distributions,
I
have
some
concerns,
especially
some
non-glipc
based
ones,
so
not
sure
about
alpine
on
response,
yet
at
least
yet
so,
but
glibc
has
distributions
like,
for
instance,
ubuntu
might
be
an
option
so
for
if,
since
is
there
is
so
so,
at
least
I
can
check
that
this
row,
but
they
asked
about
the
distribution.
D
I
said
I
use
as
that
I
use
ubuntu,
so
they
seems
to
provide
ubuntu.
So
I
can
try
to
ask
them
if
this
is
possible.
If
they
hear
us
approve
that
if
they
would
approve
that,
I
might
try
to
deploy
what
you
would
propose,
for
instance,
so
you
can
as
propose
the
chinese
engineers
to
deploy
something
they
can
try
to
have
some
fun
to
do
that
on
on
their
own,
so
you're
providing
related
tools
for
the
ci
and
it
might
be
possible
to
like.
D
There
are
two
sub
options
and
you
can
possibly
add
it
to
your
to
the
agenda.
One
option
is
to
build
everything
on
risk
5
and
then
you
are
running
all
the
tests,
integration
tests
and
some
any
other
tasks
on
the
board
itself.
Another
sub
option
is
to
cross,
compile
on
some
say,
powerful,
x86
underscore
64
server
or
maybe
that's
a
powerful
arm
server.
D
You
can
try
to
build
stuff
like
something
to
go,
perform
in
parallel
compiles
and
you
can
link
and
that
then
you
can
deploy
everything
say
run.
Then
you
are
preparing
building,
so
docker
images
and
everything
to
run
tests,
and
you
can
try
to
run
those
tests
say
on
the
board
itself.
So
the
testing
environment
could
be
risk
fire,
everything's,
five
board,
but
that
build
was
a
environment
where
you're
building
is
not
respond.
So
it's
about
cross
compilation
also,
it
could
be
possible
to
run
some
benchmarks.
D
It
might
be
about
storage,
connectors,
two
or
rise
five
boards,
and
some
plus
it
could
be.
Is
one
and
remember
run
something
in
memory
I
mean
ram
also,
it
might
be
possible,
say
to
use
some
to
run
something
like
running
some
executables
which
are
stored
completely
in
memory
files
in
memory,
and
you
can
run
some
pure
in-memory
benchmarks
to
get
some
footmen
footprint
estimations
calculations
and
you
can
get
some
understanding
on
the
on
rooks
specific,
like
functional
internal
stuff,
which
you
know
well.
D
I
guess-
and
you
can
get
some
like
insights
from
that
and
that
could
be
done
basically
in
memory.
So
this
is
one
of
the
performance
benchmarking
option,
one
of
the
performance
benchmarking
options,
so
so
ci
possible
to
to
deploy,
build
environments
possible
to
deploy
plus
you
can
combine
any
any
non-response,
xt6
underscore
64.
D
64
and
any
armed
servers,
probably
where
you
can
cross
compile
and
you
can
deploy
deploy
like
those
testing
images
on
say
and
or,
and
also
any
other
on
something
you
can
run
use
the
board
left
as
an
as
an
on
premise
and
you
can
run
any
executables
there
in
on
lending.
So
I
I
guess
at
least
such
at
least
some
suggestion.
There
is
a
host,
so
the
cost
would
be
risk
five
and
you
are
having
some
an
account
out
there.
D
So
it
could
be
done
so.
Service
could
be
operated
remotely
by
hand
and
also
they
might
pick
up.
Some
shadows
worked
workloads
and
probably
it's
possible
to
configure
those
workloads
to
actually
so
you
can
have
time
to
prepare
something
what
you
are
expecting
to
be
to
run
on
this
five
server
and
then
you
can
experiment.
It
will
just
simply
like
pick
pick
it
up
and
it
will
try
to
launch
and
you
get
get
all
the
data
from
the
output.
I
guess
it
could
be
also
possible.
A
So
maybe
a
couple
of
questions
blaine
had
some
questions.
Blaine,
do
you
want
to
ask
those
wrote
them
in
the
agenda.
B
Yeah,
okay,
I'm
off
mute.
I
was
I
I
guess,
like
my
my
first
question
is
like
rick:
is
it
like?
Rook
is
not
just
like
a
go
ago
program.
We
also
rely
on
stuff.
I
didn't
see
any
ceph
builds
for
risk
five.
B
It
doesn't
mean
that
someone
couldn't
build
ceph
for
risk
five,
but
I,
like
part
of
the
complexity
as
this
will
be
running
seth
on
risk
five,
making
sure
that,
basically,
everything
that
goes
into
that
stuff
container,
including
infestanesha,
I
guess
rook-
doesn't
make
use
of
any
of
the
iscsi
stuff
that
just
kind
of
is
present
in
that
container
can
all
can
all
build
like
it
seems
like
ubuntu
has
images
that
are
built
for
risk
five.
I
think
fedora
as
well,
so
like
the
os
itself
is
not
the
problem.
B
I
think
it
is
whether
we
could
actually
get
seth
built.
I
think,
will
be
the
most
complicated
part
of
this.
I
think
golang
provides
tooling
for
risk.
Five
like
it
sounds
like
kubernetes
has
some
you
know
experiments
for
supporting
risk.
Five.
I
think
I
think
seph
really
is
the
bottom
line
here.
B
As
far
as
can
brooke
run
on
risk.
Five
right.
A
B
I
also
like
I'm
I'm
making
the
assumption.
We
should
really
focus
on
64-bit
only
I
I
think
there
are
32-bit
risk
5
systems,
but
I
I
assume
that
that
is
those
are
pretty
low
powered
systems.
B
D
Okay,
so
I'm
as
as
far
as
I'm
such
as
seeing
what
what
you're
trying
to
to
to
like
to
to
to
two
of
you
for
us,
is
that
it's
about
64
five
watch
boards,
which
they
actually
are
plus.
I
don't
think
there
are
some
any.
D
At
least
I
haven't
heard
about
any
virtualizations
virtualization
happening
on
non-64-bit
boards,
also
plus
resource
resource
constraints.
I
don't
think
certif
beats
boards.
They
don't
have
such
legal.
Some
limitations
on
resources
so
like
so
64-bit
is
not
the
only
option.
I
guess
I
I
actually
got
some.
I
got
two
small
boards
for
s5.
D
It's
definitely
low
memory
support,
but
it's
still,
even
if
I
would
check
more
modern
options,
it
still
would
be
64
bit,
so
you
are
concerned
about
seth's
step
and
about
tests
on
risk
five
and
generally
using
sap
support
of
linux.
D
So
if
there
is
a
risk
five
horse
machine
like
and
then-
and
we
can
try-
probably
around
some
checkout-
is
possible
if
it's
possible
to
build
sev
and
runs
the
test
for
ceph
itself.
As
far
as
I'm
understanding
your
concern
so
you're
more
worried
about
file
system
supports
and
linux
for
linux,
like
kernel
models
for
that
and
some
user
space
software
for
that
matter,
right.
B
I
I
guess
it:
it's
not
specifically
limited
to
file
system.
Ceph
is
the
underlying.
B
B
B
Yeah,
I
assume
like
there
is
some
like
there's
some
precedent,
because
people
want
to
support
like
unix
or
like
bsd,
and
people
want
to
support
like
power
pc
architectures.
B
So
it
has
been
done
to
build
stuff
for
other
architectures.
But
I
I
don't
know
what
kind
of
engineering
resources
it
took
to
get
those
and
or
how
many
resources
it
takes
to
maintain
those.
D
Yeah,
okay
thanks,
so
I
think
I
have
two
language
pro
go
length,
specific
questions.
D
One
is
about
using
request,
reflection
and
the
reflect
package,
which
is
a
question
of
some
concern
in
sense
of
performance
and
the
story
with
that
jason
dash,
iterator
dependency,
which
can
be
found
against
for
kubernetes,
and
I
wanted
to
ask
if
there
is
any
plans
to
to
ch
to
investigate
the
performance
with
reflection
regarding
rook,
because
there
I
know
that
jason
dash
iterator
have
seems
to
have
some
dependency
on
alternatives.
D
I
think
I
even
opened
a
gig
git
cup
ego
easier
in
the
past
in
the
past,
so
so
it
was,
and
it
was
about
migrating
to,
like
maybe
chickens
for
some
until
alternative
alternatives,
because
reflect
true
package,
for
instance,
is
not
seem
to
be
maintained.
D
Despite
it's
supported
by
kubernetes,
we
are
just
adjacent
to
this
iterative
dependency.
That
means
that
it's
it
doesn't
use
the
reflect
package
from
the
main
library,
but
it
uses
this
alternative
and
that
alternative
not
always
is
not
always
compiled
by
an
alternative
compiler.
For
instance,
it's
it's
okay
for
gcc
go,
but
it's
not
okay
for
go
elevate
and
after
investigation
it
see
it
appeared
that
ireland
iron
taylor,
proposed
to
like
do
some
performance
benchmarking
and
provide
some
patches
to
improve
their
main
library
package.
D
Since,
if
someone's
use
this
the
reflection
library
any
in
for
concrete
concrete
needs,
say
it's
making
sense
to
optimize
like
the
main
library,
not
the,
and
to
avoid
writing
reflection
api
each
time
on
each
case.
So
there
is
no
sense
to
generate
alternative
back
ap
apis
say
it
makes
sense
to
optimize
the
reflect
package,
since
it
was
a
matter
of
some
concern
and
replacement.
D
So
I
wanted
to
ask
about
what
does
rook
here
here
still
have
on
this
dependency
for
attack
and
where
and
how
so
and
your
vision
might
be
about
that.
A
I
think
the
best
thing
might
be
to
do
if
you
could,
if
you
could
open
github
issues
for
these
two
to
discuss
there
and
point
out
specifically,
you
know,
are
you
seeing
an
issue
or
is
it
just
a
suggestion
or
an
investigation,
because
that,
because
I
mean
as
far
as
performance
I
mean
rook
is
not
on
the
data
path,
and
you
know
rook
is
the
the
on
the
management
plane.
So
just
configuring
staff,
basically
so
performance,
is
usually
not
really
a
question
for
rook
for
getting
high
performance
out
of
the
reconciling
engine
right.
A
So
if
you
see
specific
issues,
we,
I
guess
we'd
like
to
look
into
that,
and
certainly
if,
if
as
packages
are
not
supported
anymore,
we
need
to
update
the
code
to
use
the
latest
supported
packages
instead,
like
the
reflector
2,
I
think
you
said
so
that
that
makes
sense,
but
I
think
yeah,
if
you
could
whatever
details
you
have,
if
you
could
put
them
in
github
issues,
I
think
that
would
be
most
helpful
or
we
can
chat
about
it
in
slack.
Are
you
in
the
rook
slack
as
well.
D
I
used
to
be
in
the
past,
but
I
guess
I
can
jump
jump
there
and
probably
once
I
was
discussing
about
risk
five,
it's
there
is
a
slack
workspace
for
risk
five
community
as
well.
So
I
guess
there
might
be
some
sense
to
invite
some
people
there
to
like
do
some.
A
Sure
yeah
those
who
get
involved
with
getting
that
working
could
join
that
for
sure,
okay,
yeah
and
feel
free
always
to
bring
up
topics
on
slack
or
or
github.
Since
we
only
meet
here
every
two
weeks,
it's
just
a
long
time
to
wait
for
things
so
yeah.
We
can
chat
about
all
this
offline
too.