►
Description
Kubernetes Storage Special-Interest-Group (SIG) Volume Populator Design Meeting - 20 April 2021
Meeting Notes/Agenda: -
Find out more about the Storage SIG here: https://github.com/kubernetes/community/tree/master/sig-storage
C
B
You
can
see
all
right,
excellent,
okay,
so
hello,
welcome.
This
is
the
sig
storage
weekly
meeting
for
volume,
populators
I'll
I'll
start
off
with
a
status
on
where
things
are.
Last
week
we
had
talked
about
making
changes
to
some
of
our
dependencies
to
try
to
I've
been.
You
know,
responding
to
code
review
comments.
B
A
lot
of
the
code
review
comments
were
updating
versions
of
stuff,
and
so
we
we
went
in
deep
last
week
on
exactly
how
we're
building
a
lot
of
the
auto
generated
code,
because
that
has
some
awkward
dependency
structures
and
so
what
I've
done
is
removed
all
of
the
dependencies
on
kate's
dot,
io,
slash
code
generator
and
now
we're
only
using
controller
gen
from
the
q
builder
project,
and
so
what
that
means
is
there
are
no
more
statically
generated,
clients
or
client
sets
or
informers
or
listers.
B
The
implementation
now
uses
the
dynamic,
the
dynamic
client
with
a
little
bit
of
syntactic
sugar
on
top
to
convert
to
the
appropriate
types
of
objects,
which
is
what
I
understand,
q
builder
does
internally.
I
think
q
builder
relies
exclusively
on
the
dynamic
client
and
then
just
puts
syntactic
sugar
on
top.
So
that
you
can
get
typed
interfaces,
you
know
at
the
level
that
you're
writing
your
logic.
B
The
main
thing
is
both
of
these
projects
are
so
small
in
terms
of
lines
of
code
that
it's
not
clear
that
q
builder
would
make
them
any
smaller.
It
might
make
them
a
lot
larger,
and
that
would
not
be
good.
So
my
my
cost
benefit
analysis
is
you
know,
cubita
is
a
win
only
if
it
makes
things
smaller
and
simpler,
rather
than
larger
and
so
far
the
switching
the
dynamic
client
definitely
shrunk
things.
So
that
was
good
yeah.
B
Okay,
any
questions
about
that.
Anyone
want
to
see
any
of
that
code.
I
haven't
actually
pushed
the
pr
yet
so
it's
still
sitting
in
a
repo
on
my
machine.
I
was
testing
it
as
recently
as
yesterday
to
make
sure
that
I
didn't
break
anything
and
it
seems
to
be
solid,
so
I
can
safely
push
it.
I
just
didn't
have
time
to
push
it
this
morning,.
B
All
right,
if
everyone's
cool
with
that
that
sounds
good,
I
don't
suppose
assad
you
were
ever
able
to
review
these
two
repos
and
the
code
in
them.
I
have
not
now,
okay
shane
did
you
review
the
other
repo,
the.
C
B
Source
validator
controller,
but
yeah,
I'm
still
waiting
for
code
reviews
on
the
other
ones.
Okay,.
B
I'm
gonna
push
towards
you
know
getting
them
in
shape
with
prow
and
doing
and
release
tools
and
doing
the
releases
as
soon
as
possible.
It's
just
week
by
week.
At
this
point,
when
I
find
time.
The
other
thing
I
wanted
to
talk
about
is,
as
we
go
into
122,
that
the
plan
is
to
move
to
beta,
which
means
merging
my
updated
cap,
which
is
still
floating
out
there.
I
probably
need
to
refresh
it
or
rebase
it
and
getting
the
approvals
from
tim
and
whoever
else
so.
B
Well,
I
mean
it's
nice
in
that
it's
it's
a
more
luxurious
schedule
for
developers,
but
it
ultimately
means
you
get
less
chances
to
push
code
out
every
year.
You
know
every
time
you
make
a
release
slower.
You.
C
B
Yeah
I
mean
at
the
maturity
level.
Kubernetes
is
that
slower
might
be
better,
so
so
yeah,
I
just
I.
I
don't
want
to
wait
until
like
we're
staring
at
the
deadline
for
caps
again
to
to
try
to
get
the
cap
merged
in
122..
I
want
to
be
well
ahead
of
the
schedule,
so
I
want
to
make
sure
we
have
an
eye
on
when
the
deadline
actually
is
and
when
we
can.
I.
C
B
Yeah,
I
figure
that's
the
case,
so
I
just
have
to
do
my
side
of
the
work
which
is
to
rebase
the
cap,
updated
with
any
changes
that
have
happened
in
the
last
release
and
then
send
out
requests
for
review
again.
B
One
yeah,
so
so
what
happened
in
122
is
is
we
had
one
kept
and
tim
took
a
look
at
it
and
he
said
yeah
this.
This
seems
fine,
but
there
was
a
a
proposal
in
my
kept
to
sort
of
break
backwards,
compatibility
in
one
very
strange
corner
case,
and
he
and
he
suggested
that
we
break
that
out
into
a
separate
cap
and
consider
it
independently,
and
so
I
have
a
a
pr
against
enhancements
that
to
split
the
one
kept
into
two
cups
and
that
has
not
been
merged
yet.
B
So
that's
what
needs
to
happen
and
maybe
in
maybe
in
the
next
one
of
these
meetings,
we'll
go
over
that
kept
in
this
group
and
and
then
work
on
getting
the
api
reviewers
in
to
review
it.
If
I
can
just
get
this,
I
just
want
to
get
the
the
alpha
code
released
first,
so
that's
like
my
top
priority.
That's
what
I'm
driving
towards.
B
So
I
wanted
to
mention
that
yeah,
the
122
beta
planning,
okay
yeah,
those
were
the
main
items
on
on
my
my
agenda.
I
know
we
have
some
new
people
who
poked
me
on
the
the
public
slack
channel,
welcome
chauvin
and
ashitosh.
Did
you
guys
have
a
specific
topic?
You
wanted
to
discuss
or
questions
you
wanted
to
ask,
or
are
you
interested
in
contributing.
D
Hi
ben
ashtosh
here
just
a
quick
introduction,
I'm
based
out
in
bangalore,
india,
and
you
know
I've
been
contributing
to
open
ebs,
nothing
specific
top
of
my
mind.
I
was
just
you
know,
experimenting
with
this
volume
populator
and
I
thought
you
know
if
there
is
anything
in
terms
of
code
or
documentation,
I
would
be
happy
to
contribute
there.
So,
oh.
B
Okay,
excellent
yeah
yeah
the
whole
point
of
releasing
an
alpha
is
for
people
to
actually
try
it
out
and
tell
me
why
it's
why
it's
broken,
and
so,
if
you
haven't,
had
problems
yet
that
that's
a
good
sign
but
yeah
the
moment,
you
have
any
issues
with
anything
file
issues
or
you
know
ping
me
on
slack
or
come
to
the
meetings
and
yeah.
If
you
want
to
maybe
help
with
I'm
thinking
about
your
documentation,
we
need
some
readme
files
that
are
very
clear
in
these
repos
that
explain
how
to
how
to
use
it.
B
We
need
one
of
the
repos
is
completely
lacking
unit
tests.
I
gotta
add
unit
tests
and
I'm
forgetting
which
one
it
is.
I
think
it
might
be.
The
the
data
source
validator
needs
unit
tests
and,
if
you
know
anything
at
all
about
csi
release
tools,
you
can
come
help
me
because
that's
that's
my
biggest
pain
point
right
now
is.
I
haven't
been
able
to
talk
to
patrick
yet
about
why
I
can't
get
release
tools
to
behave
properly,
but
that's
that's
the
next
thing
on
my
agenda.
After.
C
Ben
you!
Actually,
if
you,
if
you
submit
some,
let's
say
working
progress
pr
with
the
problems,
then
you
just
assign
that
to
patrick
asking
you
to
take
a
look.
I
think
he
will
take
a
look
if
you
pinned
him
there
on
your
pr.
C
B
Yeah,
that's
a
good
idea
yeah,
given
that
we
don't
have
any
sort
of
automated
builds
with
pro.
You
know,
pr's
that
get
pushed
nothing
happens.
Oh
so.
B
C
Oh,
but
you
can
just
yeah,
so
you
can
just
like
so
so
you're
saying
you
can,
if
you
just,
if
you
just
use
the
the
instructions
right,
you
basically
get
all
the
up
latest
update
from
release
stores
and
then
you're
saying
the
problem
is
what
you
cannot
and.
C
C
B
Yeah
yeah
morning
time
would
be
a
good
time
to
reach
them.
Okay,
yeah.
I
like
those
suggestions.
I
just
I
need
to
spend
more
time
actually
working
on
the
release
tools
piece.
I
I
spent
a
lot
of
last
week
doing
all
the
conversion
from
kate
said:
io
code
generator
to
controller
gen.
So
now
all
the
deep
copy
functions
are
being
generated
by
controller
gen
instead
of
the
code
generator
and
then
removing
the
dependency.
B
So
I
wanted
to.
I
got
the
impression
from
either
I
don't
know
if
it
was
ashitosh
or
chauvin
that
one
of
you
was
actually
trying
to
implement
a
populator
of
some
kind.
Did
you
want
to
talk
at
all
about
what
what
you
had
in
mind
for
a
use
case.
D
Yeah,
I
think
someone
had
done
the
implementation
and
he
was
demoing
and
it
was
an
awesome
demo.
So
someone
if
you
want
to
talk
about
that-
and
I
also
do
have
a
use
case-
I
can
say
after
so.
A
Hello:
everyone,
okay,
so
open
abs
contributor.
So
I
was
exploring
this
feature
for
open
eps
use
cases
and
I
was
actually
demonstrating
this
feature
in
openbps
community
meeting.
A
So
in
the
in
that
demonstration
I
was
actually
building
a
react
application
in
my
local
and
then
pushed
that
application
to
s3
bucket
and
I
was
creating
the
volumes
from
that
s3
data
source
and
I
was
just
scaling
up
the
stateful
set
yeah.
That
was
the
demonstration
I
have
done
in
in
a
community
called
open
abs.
B
A
B
Cool
yeah,
yeah,
we'd
love
to
have
links
to
you,
know
things
that
are
actually
using
this
to
sort
of
show
that
you
know,
there's
real
demand
here
that
was
part
of
our
graduation
to
beta
requirement
in
the
cap
is
we
want
to.
B
D
Right
and
the
use
case
that
I
have
been
looking
into
then,
is
like
you
know:
we
are
deploying
in
and
postgres
and
click
house
database
and
these
two
databases
actually
when
they
start
up
for
the
first
time,
and
they
actually
need
a
lot
of
pre-filled
data
to
be
pushed
into
the
disk.
So
it
is
just
some
cve
notification
stuff.
D
You
know
that
the
back
end
used
to
generate
some
cv
alerts,
and
what
we
have
observed
is
that
this
data
has
grown
to
use
amounts
and
we
don't
want
to
actually
start
the
database
and
then
pull
from
internet
and
then
just
post
the
stuff.
So
I
think
it's
interesting
I'll.
Try
that
out.
How
can
it
help
to
push
the
data
via
the
dump
files?
So
that
is
something
that
I'll
try
out
and
get
back
to.
You.
D
We
have
some
jump
file,
that's
stored
somewhere
else,
and
we
want
to
just
see
if
you
know
that
dump
can
be
used
to
like
we
don't
want
to
go
by
the
init
containers
way
somehow
and
so
we'll
we're
just
looking.
We
were
just
exploring
how
we
could
actually
do
something
that
you
know
it
previously
comes
up
with
some
pre-filled
data,
and
then
it
just
goes
hunky
dory.
So
that's
the
plan.
Okay,.
B
Yeah
I
mean
that
that's
exactly
the
kind
of
thing
the
population
lets
you
do
right,
I'm
curious
what
the
difference
is
between
that
and
like
just
having
a
a
pvc
that
already
contained
your
data
and
you
cloned
the
pvc.
D
B
Yeah
yeah
and
if
you
have
any
any
links
to
to
implementations
or
anything
post
them
in
the
agenda,
so
we
have,
we
can
keep
track
of
them
off
the
agenda.
B
That
would
be
great
sure.
B
Thank
you
yeah
all
right,
so
that's
it
for
for
my
agenda.
If
nobody
has
any
questions
or
anything
else,
I
will
get
back
to
hacking
and
pushing
my
pr.
I
think
it's
it's
basically
ready
to
go.
D
B
Oh,
oh
yeah
yeah,
it's
just!
I
just
never
got
around
to
adding
the
unit
tests
in
in
the
version
that
made
it
into
the
initial
commit
so
yeah.
If
you,
if
you
wanted
to
add,
like
a
test
file
and
we
might
have
to
refactor
the
code
a
little
bit
for
testability,
but
I'd
like
to
have
something
to
be
able
to
run
in
the
proud
job
when
feature
changes
come
in
to
detect
regressions
and
then
we'll
expand
from
there
on
on
testing
we'll.
Thank
you
on
slack.