►
From YouTube: Kubernetes SIG Node 20200728
Description
Meeting Agenda:
https://docs.google.com/document/d/1j3vrG6BgE0hUDs2e-1ZUegKN4W4Adb1B6oJ6j-4kyPU
A
With
the
summary
and
the
motivation
part
of
the
cap,
I'm
wearing
some
reviews
on
that
it
should
be
quite
straightforward
because
it's
based
on
all
their
proposals.
So
I
think
there
is
no
there.
There
should
not
be
a
big
discussion
here.
I
have
a
question,
I'm
also
working
on
the
proposal
part.
So
should
I
wait
for
this
first
pair
to
be
merged,
or
can
I
open
another
pr
with
the
proposal
and
design
details
right
now.
B
We
normally
prefer
working
on
existing
evolve
existing
of
the
crab,
but
I
know
this
this
situation
a
little
bit
different,
so
dark
because
there
used
to
be
a
red
height,
actually
driving
the
username
space.
And
how
do
you
prefer
to
evolve
from
here?.
D
For
more
specific
use
cases,
I
don't
know
if
we
had
anything
we're
ready
to
share
externally
on
what
the
thoughts
on
that
would
have
been.
But
I
think
what
we
were
starting
to
think
through
is
like.
Did
the
original
phased
approach
on
doing
node
level
user
namespace
remapping?
D
Is
that
necessarily
needing
to
be
the
first
step,
or
is
it
possible
that
we
could
get
better
clarity
on
the
actual
use
cases?
Folks
are
trying
to
support
to
see
if
we
could
get
more
targeted
or
granular
solutions?
So.
D
Maybe
in
this
case,
what
we
could
do
on
the
cap
is
see
if,
in
the
broader
community,
we
can
share
what
the
use
cases
are
and
if
there
are
use
cases
that
are
more,
let's
say,
like
micro,
targeted
so,
for
example,
on
the
right
outside
we're
very
interested
in
getting
like
the
ability
to
unprivileged
container
builds
and
in
many
of
those
environments
like
you,
don't
need
to
care
about
pvc
remapping,
and
it
felt
like
there's
opportunities
for,
like
smaller,
more
targeted
solutions
that
one
could
do
without
needing
to
do
everything
for
that
type
of
use
case,
and
so
I'm
curious,
if
there
are
others
who
might
just
want
to
be
better,
might
want
to
more
transparently
speak
about
what
targeted
use
case
scenarios
username
space
remapping
is
interesting
for
or
if
everyone
feels
like
they
wanted.
D
E
Yeah,
I
I
think
that
you
covered
pretty
much
everything.
I
think
what
we
can
do
is
disappear.
I
can,
and
I
can
work
together
on
the
exact
use
case
and
how
we
think
we
can
solve
it
and
maybe
present
in
a
week
or
two.
D
I'd
like
to
get
the
cases
better
fleshed
out
and
we
have
more
recent
performance
data
on
at
least
what
we've
measured
with
respect
to
namespace
remapping
than
when
we
started
to
explore
this
two
years
ago.
So
I
think
the
near
next
step,
I'd
like
to
see
us
do
is
just
like
s:
community,
better
quantify
the
near-term
use
cases
we're
looking
to
solve,
and
I
don't
think
we
did
a
great
job
on
that
in
the
initial.
G
B
His
new
cap
about
this
and
how
to
improve
so
maybe
we
can
start
from
that
from
his
new
cap
since
he
already
shared
and
published
so
then
we
can
decide
emerge
altogether.
I
I
noticed
that
there
are
some
small
use
cases,
but
I
haven't
looked
at.
I
didn't
read
his
proposal
in
detail,
so
not
not
sure
he'd
address
any
problem.
Like
a
measure.
Do
you
want
to.
A
Yes,
that's
fine
for
me!
Yes,
actually,
I
was
thinking
on
also
opening
the
pr
about
the
proposal,
because
what
I
did
was
to
take
the
old
proposals
and
more
or
less
combine
them
together
to
create
something
like
a
new
proposal,
and
so
yes,
my
idea
was
to
open
the
and
I
started
this-
this
discussion
there.
So
I
don't
know
if
we
want
to
go
back
a
bit
and
discuss
the
use.
The
use
cases
again.
D
So
this
is
the
first
time
seeing
the
proposal.
That's
open.
So
apologies
for
that,
but
maybe
it's
a
there's
a
clear
next
step
like
so.
We
have
a
sufficient
body
folks
working
around
it.
I
see
you
see
seed
bernal
on
the
proposal,
maybe
if
we
could
just
get
renault
and
giuseppe
to
share
their
insight.
That's
a
clear
next
step.
E
Yep
that
sounds
good
I'll,
make
a
pass
this
week
and
you
coming.
E
Guess
we'll
just
leave
review
comments
on
the
existing
pr
and
I
think
it
may
make
sense
like
if
you're
updating
the
cab
right,
like
I
wouldn't
wait
for
parts
of
it
to
be
merged.
Just
add
everything
to
the
pr
I
see
a
like.
Some
sections
aren't
filled
out
yet
right.
So
if
you
have
anything
else
in
the
pipeline,
I
would
suggest
just
update
the
your
pr
already
for
that.
Oh.
A
D
Yeah,
I
guess
yeah
like
the
quick
pass
on
the
cap
is
like
to
me,
and
I
repeat,
this
is
just
like
persistent
volume
management
was
the
hardest
part
of
doing
this
well,
and
so
there
are
things
I
think
we
could
do
at
the
container
runtime
level
without
needing
for
use
cases
that
didn't
need
to
support
persistent
volume,
and
so
I
guess
in
general,
I'm
wondering
is
like
the
author
on
this
cap
is
your
use
case
requiring
consumption
of
pvs.
A
D
Okay,
so
I
think
then
it'd
be
great
for
you
and
criminal
just
happy
to
get
together
on
that,
because,
if
you
weren't
trying
to
tackle
pvs,
I
know
we
here
at
the
right
outside
had
some
ideas
on
how
we
can
handle
this
for
cases
and,
if
you're
looking
to
do
things
with
like
techton
or
anything
else.
That's
doing
builds
on
top
of
cube
as
a
use
case,
and
you
wanted
to
do
it.
You
know
rootless
or
something.
D
D
C
E
Action
to
set
something
up-
I've
already
single
directly
in
the
past,
so
I'll
ping.
Him
then.
G
Yeah,
that
would
be
awesome
because
also
like
we
have
several
use
cases
like
security
improvements.
There
are
use
cases
where
we
don't
need
pvs
and
the
security
improvements
this
gives
use
are
way
better,
so
yeah.
I
think
there
are
some
several
use
cases
when
we
don't
need
to
solve
the
whole
big
problem.
Reducing
the
scope
might
be
might
be
actually
very
helpful.
B
Okay,
thanks
looking
forward,
you
come
back
to
signal
and
report
with
the
agreement,
and
hopefully
we
can
have
the
presentation
here
and
the
next
move
to
me
said
ordering.
G
So
yeah,
I'm
with
the
show
joseph
everyone
we're
finishing
the
pr
with
all
the
call-outs
and
different
alternatives
and
suggestions
for
for
yeah
for
all
the
for
all
the
edge
cases
we
detected
it's
written
in
a
quite
it's
it's
quite
long,
so
we
try
to
err
on
the
size
of
being
long
but
very
clear
about
what
the
collab
is
but
yeah.
G
So
I
wanted
to
warn
you
on
that,
but
once
we
decide
on
one
alternative,
we
can
basically
remove
that
or
move
it
as
a
summary
in
in
the
alternative
section
that
were
considered
yeah.
So
it
shouldn't
be
such
a
big
problem
like
the
proposal.
Part
should
still
be
sustained
and
clear.
G
H
Hey
rodrigo,
it's
okay,
I'm
sorry!
I
dropped
the
ball
a
little
bit
on
that
which
pr
is
currently
active
for
no,
no.
G
No
drop
the
ball.
I
still
didn't
open
the
vr,
but
I'm
planning
to
open
a
pr
with
all
these
callouts.
It
will
be
quite
long
because
all
the
callouts
are
explained
in
details
and
what
are
all
the
implications
that
this
decisions
in
the
current
cap
has,
but
and
for
every
issue
that
we
found.
G
We
propose
several
alternatives
and
a
suggestion,
and
what
I
was
trying
to
say
just
is
that
this
is
quite
long,
but
once
we
agree
on
an
alternative
on
how
to
address
it,
this
can
all
be
removed
and
just
some
summary
of
the
discussion
be
capturing
the
alternative
section
and
just
modify
the
proposal
accordingly
to
what
we
decided.
G
G
Also.
I
wanted
to
mention
that
next
week,
I'll
be
off
for
maybe
two
or
three
days
so
yeah.
If
I
don't
answer
I'll,
be
travelling
back
home
because
due
to
chlorina
virus,
I'm
I'm
stall
in
in
in
one
country,
but
I
I'm
not
at
home
next
week
I'll
be
able
to
go
home
so
yeah
next
week.
I
might
be
off
for
two
or
maybe
three
days.
B
There
you
go,
I
hope
you
have
safe
trip
back
at
the
price,
your
treatment.
B
B
Okay,
the
gems
do
we
thank
you
for
came
to
the
signal.
I
Hello,
so
my
name
is
james:
I'm
on
the
q
nets,
119
release
team
working
with
release,
notes,
119
is
scheduled
to
release
in
just
under
a
month
and
we're
drafting
the
final
release.
Notes
now
we're
going
around
and
asking
every
sig
what
they
feel
their
major
themes
are
of
the
release.
So
I'd
be
interested
to
hear
what
people
have
to
think
about
that.
I
can
take
suggestions
now
or
as
a
discussion
or
I
can.
B
D
Yeah,
so
I
think
james
I'm
happy
to
if
you
pick
me
on
slack,
I'm
happy
to
help
put
together.
D
The
themes
I
think
stuff
that
would
stand
out
to
me
is
like,
where
we've
made
progress
on
kind
of
early
phase
features.
So
the
secrets
v2
work,
made
a
lot
of
progress.
There
were
some
enhancements
around
topology
management
support
that
we
highlighted
in
the
prior.
D
So
if
we
wanted
to
work,
we
could
talk
through
that,
but
I
think,
what's
best
james.
If
it's
okay,
I
could
send
you
maybe
three
or
four
sentences.
I
Later
this
week,
yeah,
that's
completely
fine,
there's
no
enormous
rush.
Here,
we've
got
about
a
month
left,
as
I
say
so,
just
just
want
to
get
people
thinking
about
it,
really
yeah,
no
problem
cool
right.
Thank
you
very
much.
B
D
Guess
that's
how
I
would
would
look
at
it.
I
guess
the
testability
stuff
that
victor
helpdrive,
I
think,
is
a
good
thing
to
also
call
out.
B
I
think
there's
the
yeah
accelerator,
like
the
what
the
the
the
the
deceiver
of
accelerator,
monitor
monitoring
metrics
that
could
be
called,
because
that's
the
one
way
we
we
definitely
want
to
meet
the
community.
No
and
there's
in
the
chat.
There
are
people
mentioning
that.
That's
that
definitely
said
comp
ga
graduation
worse
to
medicine,
yes
yeah,
so
so
so
we
definitely
have
some
like
the
graduation.
We
definitely
need
to
mention
I'm
just
concerned
about
the
new
feature.
If
we
don't
have
enough
tests,
we
even
don't
have
tests
on
the
community.
Then
we
call
out.
B
So
that's
my
concept,
because
we've
been
discussed
this
for
windows
long
time
back
right.
So
we
said:
if
there's
no
public
test,
we
cannot
really
sponsor
or
support.
Even
we
have
seek
windows
to
to
put
a
test
there,
but
if
there's
no
test
and
it's
hard
for
us
to
justify
those
things
so
anyway,
I
just
share
my
thoughts
here.
B
Thanks
the
assessor
about
to
to
to
volunteer
to
write
something
up
about
the
second
year.
D
Yeah
I
had
one
topic
if
it's
okay
dawn,
one
of
the
things
that
I've
been
meaning
to
raise,
but
I
haven't
had
a
good
forum.
It's
on
my
mind.
Now
we
have
time
is
lori
apple
had
been
reaching
out
talking
about
if
we
wanted
to
start
running,
maybe
a
dedicated
triage
meeting
to
help
get
priority
focus
on
kind
of
unblocking
of
prqs.
D
Given
that,
like
we
just
elevated
david
and
seth,
to
maintain
our
roles,
it
felt
like
we
were
kind
of
backfilling
holes
that
we
had
where
maybe
we
could
more
effectively
start
to
run
some
of
these
triage
meetings
with
like
with
some
coverage
of
ensuring
that
there
wasn't
a
an
approver
in
any
form
we
were
holding.
D
So
what
I
was
curious
about
throwing
out
here
is
like
if
we
were
to
put
something
like
that
together
at
what
cadence
would
we
think
it
would
be
helpful
and
if
any
of
the
members
of
this
tech
have
maybe
joined
other
cigs
triage
meetings
to
see
what's
valuable,
that
we
could
go
and
look
at
role
modeling
before
kicking
it
off
ourselves.
H
So
before
before
we
it's
sergey
speaking,
I
was
thinking
about
that
that
I
can
I'm
writing
this
query
right
now.
It's
you
know,
music
notes,
so
I'm
trying
to
understand
how
many
pr's
are
currently
in
signaled
area,
and
it's
surprisingly
that
most
of
the
prs
is
just
a
cleanup.
So
I
think
cleanup
and
box
are
two
are
the
most
prominent
areas.
I
unfortunately
cannot
share
my
screen
because,
I'm
speaking
from
a
phone,
but
if
you
can
click
on
this
link,
it
may
be
useful.
I.
J
G
H
H
And
yeah-
and
I
did
this-
I
did
a
spreadsheet
of
burn
down
of
pull
requests.
I
mean
I
started
tracking
it
manually
before,
but
then
I
started
writing
them
down
and
I
wanted
to
understand
what
working
us
and
that's
why
I
sent
this
pull
request
to
clean
up
owner's
file
and
at
david
and
sas
to
more
on
your
styles.
H
So
I
think
don
approved
this
pull
request
and
I
also
noticed
that
the
sample
requests
are
not
blocked
by
us
is
blocked
by
hack
team
owners
for
cleanup,
especially
so
I
send
this
pull
request.
So
if
you
can
watch
for
this
pull
request
as
well,
it
will
be
helpful.
I
added
into
notes.
So
if
you
scroll
all
the
way
to
the
right,
there
is
a
turn
down.
H
H
So
there
is
a
number
total
number
of
pull
requests
and
breakdown
by
if
you
just
separate
break
down
by
kinds
yeah,
I.
K
H
So
yeah,
you
notice
that
kind,
bark
and
kind
clean
up
is
kind
of
the
biggest
area
and
the
worst
part
is
that
kind.
Cleanup
is
so
big
because
typically
kind
cleanup
doesn't
require
any
feature,
knowledge
or
any
I
mean
it's
supposedly
just
a
refactoring
or
some
code.
Improvements
like
linter
issues,
improvements,
something
like
that.
H
So
unfortunately,
those
pr's
got
stuck
quite
often
so
we
have
one
from
like,
I
think
more
than
a
year
and
a
half
ago
that
was
approved
by
everybody
except
hack
team,
and
then
it
was
stale
and
rotten
because
more
changes
was
accumulated
in
the
master
branch,
and
now
it's
abandoned.
H
It's
still
open,
but
it's
abandoned.
So
I
I
suggest
this
another
pull
request
to
change
hack
files
permissions,
so
every
approver
has
a
permission
to
approve
those
kind
of
pull
requests,
so
it's
kind
of
some
ideas
how
to
improve
how
to
burn
down
some
of
this
pull
request.
But
obviously
there
are
more
and
derek.
I
saw
your
comment
on
one
of
the
pull
requests
where
there
was
a
clean
up
of
some
refactoring
of
node
status
class,
and
I
totally
support
this
comment.
H
Saying
like
some
cleanups
are
meant
to
be
like
just
the
refactoring,
for
the
sake
of
refactoring,
without
clear
justification,
why
we
need
those
cleanups
and
those
cleanups
may
take
a
while
to
review
when
they
don't
bring
any
obvious
values.
So
perhaps
for
that
kind
of
pr
that
we
reject,
we
can
explicitly
reject
them
and
close
them,
so
they
wouldn't
hang
active
if
you
can
like
just
close
them
with
a
appropriate
messaging.
H
Maybe
it
will
be
okay
for
people
and
we
can
suggest
to
break
it
down
into
smaller,
prs
and
more
manageable
pieces
that
can
be
reviewed
quite
easily
and
they
wouldn't
cause
any
regression
in
the
future.
Yeah.
C
So
I
guess,
if
it's
okay,
I
I
very
much
appreciate
you
sharing.
D
These
stats
so
handily.
I
think
this
would
be
helpful
data
for
us
to
review
and
prioritize
what
I
was
trying
to
think
through
is
and-
and
I
wanted
to
sample
some
other
teams
as
triage
meetings
or
other
other
six
trials
meetings
was
like.
I,
I
wanted
to
make
sure
that
we
were
having
constant
focus
on
cherry
picks,
given
that
they
are
largely
probably
impacting
actual
users
in
production
and
that
we
didn't
get
them
lost
and.
D
Between
kinebug
and
maybe
kind
documentation
or
something
we
could
probably
prune
these
in
some
priority
order,
but
this
would
be
a
nice
banning
way
to
try
to
make
the
triage
meetings
itself
effective.
So
yeah
like
the
goal,
would
be
like
to
figure
out
what
the
right
status
quo
here
is.
I
don't
know
if
the
number
is
ever
going
to
go
down
below,
say,
150
or
something
because
many
prs
span
functional
areas,
but
I
felt
like
we
could
do
when
we
have,
but
now
and
given
some
new
energy.
D
D
D
B
Nice,
that's
good
yeah
yeah.
I
think
the
tragic
meeting
is
review
yeah,
it's
very
so.
I
also
agree
with
the
direct
priority.
I
think
the
top
one
it
is
the
cherokee
that's
approved.
It
is
the
production
issue
and
already
everyone
already
provide
the
abundant
review
and
then
we
need
to
make
sure
the
release.
The
version
have
those
cherry
pick
and
another
thing
it
is.
I
think
that
I
also
agree
with
you
like
the
kind
bug
and
kind
of
documentation,
but
also
I
want
to
add
one
more.
B
The
thinning
test,
and
especially
for
those
conform,
related
feeding
types
I
think
is
deserves
some
like
the
higher
priority
attention
from
the
charge
meeting,
maybe
and
unless
we
all
agree
separate.
The
meeting
continue
like
that's
the
other
thing
I
think,
like
the
design
and
the
and
also
feature
actually
literally
came
to
the
signal,
and
oh
maybe
off
offline
design.
Meeting
we've
been
called
out
like
a
recently
have
like
the
resource
management
related
design
meeting.
B
So
that's
kind
of
big
take
care
so,
but
the
circuit
also
mentioned
is
one
things
like
the
clean
up,
and
I
guess
that
is
kind
of
like
the
base
and
the
product
is
a
little
bit
full
into
crack.
So
we
can
continue
through
the
discussing,
like
the
slack,
how
we
are
going
to
to
pick
up
those
coming
up
things,
because
some
of
those
cleaning
up
may
help
us
maintain
off
our
code,
readability
and
maintainability.
B
H
Yeah
and
to
be
clear,
my
goal
with
kind,
cleanup
and
kind
bug
is
not
to
prioritize
them
very
high
and
pay
more
attention
on
that.
It's
just
to
streamline
the
process
with
them
and
make
sure
they
are
not
spending
time
on
them.
So
I
think,
taking
those
like
this
volume
out
of
everybody's
time
and
concentrating
approvals
time
on
future
work
might
be
very
beneficial.
So
that
was
my
biggest
intention,
but
yeah,
and
also
newcomers
like
new
contributors
to
kubernetes
have
some
hard
time
understanding
how
how
it
works.
F
Hey
this
is
victor.
I
wanted
to
just
add
one
one.
One
thought
you
know
when
we
were
doing
the
in
the
past
some
of
the
cleanup
for
the
sick
testing
stuff.
One
of
the
constant
challenges
was,
you
know
all
the
emails
coming
in
about
hey
review
this
pr
and
it
was.
J
F
To
lose
track
of
that
stuff,
and
so
what
I
did
is
looked
around
and
actually
github
has
a
project
management
tool.
I'll
just
call
it
that,
because
that's
what
it
is,
it's
a
board
and
with
that
you
can
track
all
of
the
issues,
and
so
what
I
did
is
created
a
github
board
and
I
was
only
tracking
the
sick
test
off
and
but,
for
example,
like
the
spreadsheet.
Here
I
mean
you
can
sort
this
thing
based
on
the
kind,
whether
it's
a
cleanup
or
test
or
feature.
F
I
mean
I've
got
the
board
up.
If
you
just
want
to
take
a
quick
look
or
folks
are
familiar
with
it.
That's
fine!
F
If
you
don't
want
to
be
bothered
with
that,
but
what
I'm
saying
is
there's
the
tool
that
at
least
helped
me
sort
of
get
things
pushed
through
from
the
sick
testing
cleanup,
and
you
know
that
has
been
sort
of
put
on
hold
for
now
for
a
couple
of
reasons,
but
I
think
that
would
be
a
great
tool
to
consider
to
use
for
the
triage
and
to
prioritize,
because
once
it
gets
to
a
certain
point,
you
say:
hey,
these
pr's
are
they're
all
done
all
the
tests
are
passing
and
we're
waiting
for
a
reviewer.
F
So
you
can
have
a
column
that
says
I
just
need
somebody
with
approval
rights
to
go
and
say:
okay,
make
sure,
everything's,
good
click
and
it's
it
makes
it
easier.
Especially
for
you
know,
I
imagine
don
you
and
derek
just
get
pinged
all
the
time
he
approved.
J
F
J
F
Yep
in
the
chat,
it's
there's
the
current
board
there.
If
you
just
want
to,
I
mean
you,
can
pull
it
up
and
you
can
see
it's
it's.
It's
really
cool
cool.
D
Yeah,
maybe
you
could
reach
out
to
daniel
I'll
reach
out
to
david
or
stefan
and
see
how
they're
handling
that,
but
in
general
I
thought
like.
D
With
more
rights
being
offered
to
more
people
in
the
seg,
it
was
a
good
time
to
to
try
to
fix
some
some
gaps
that
we
had.
So
let
me
it
sounds
like
there's
interest
in
this,
and
so
it's
worth
pursuing,
maybe
as
an
action
item.
Just
the
list
of
current
approvers
will
sync
up
to
see
what
the
right
cadence
is
and
then
we'll
try
to
get
a
dude
we'll
send
out
to
the
mailing
list,
with
some
proposed
times
that
match
that
cadence
to
figure
out.
D
If
we
can
get
broader
participation.
H
Cool
yeah
and
happy
to
participate-
and
I
have
one
question
just
very
tactical:
do
we
need
to
cherry
pick
orgmiers
files
to
earlier
releases
as
well,
so
you
have
approvers
that
were
added
recently,
also
being
approvers
for
older
branches.
D
So
I
would
find
that
definitely
useful,
so
yeah
I
would
ask
seth
and
david:
can
you
make
sure
you're
in
cherry
picked
versions
of
currently
supported
releases
and
we'll
prioritize
getting
those
approved,
yep
and
then
david
and
seth?
I'm
also
curious.
We
had
a
gap
in
the
testing
repos,
which
victor
was
hitting.
F
D
So
I
I
know
we
had
I'd,
think
that
before
you
all
became
approved
to
see,
if
we
need
to
ensure
we
don't
have
a
gap
on
that.
L
L
H
Yeah
test
intro
was
weird
because
I
mean
not,
I
I
don't
know
history,
but
they
were
like
derek.
Was
there
dawn?
Wasn't
there,
so
I
wasn't
sure
what
the
status
of
it
and
whether
we
like
whether
it
was
intended
for
all
kubrick
owners
to
own
it.
K
D
Okay,
well,
thanks
for
it's
so
rare,
we
have
some
time
not
discussing
very
complicated
topics
to
bring
up
some
of
the
mundane,
but
I
think
this
is
important,
so.
B
Dark
as
a
follow-up,
I
also
want
last
week
you
are
not
here,
so
we
talk
about
the
continuum
of
the
resource
management
discussing
because
it's
gone
from
from
the
so
there's
the
topic
proposed
and
also
there's
the
more
people
want
to
continue
discussing
about
the
topology
management
and
also
continue
off
like
improve
off
the
memory
as
the
resource
and
the
pneuma
a
while
scheduling
and
there's
also
have
the
people
sending
me.
There's
also
have
the
open
pr
talking
about
the
disk
io.
B
I
think
this
is
been
talked
about
a
long
time
back,
but
we
don't
have
any
concrete.
This
result
come
out
and
they
also
have
like
the
new
proposal
came
from
the
microcosm,
the
nri
interface
and
I
shared
the
michael
about
some
old
discussing
meeting
notes
and
also
the
resource
class
class,
which
is
touch
base
a
little
bit.
That
topic,
but
not
really,
we
agree
so
so
as
a
follow-up.
B
D
Yeah,
so
I
guess,
like
I
said
I
was
out,
I
know
it's
sometimes
hard
in
open
source
projects
to
realize
victor
who
was
doing
such
a
nice
job,
helping
coordinate
that
had
also
had
some
situations,
so
I'd
like
to
just
be
able
to
have
time
to
watch
last
week's
recording
and
see
the
latest
stuff
on
the
node
resource
interface.
D
But
then
I
don't
object
to
continuing
it,
but
I
I
just
let
me
follow
up
to
see
if
we
have
a
capacity
to
do
it
well
and
if
not,
then
others
should
take
it
on.
D
D
D
So
when
some
folks
from
say
six
scheduling
are
asking
for
apology,
aware
scheduling
it's
basically
a
strong
desire
for
users
who
who
need
those
needs
to
have
cluster
level
visibility
that
something
like
the
cri
would
not
make
or
sorry
the
nri
would
not
make
knowable,
and
so
it
just
felt
like
there's
this
consistent
tension
in
cube
where
it's
like.
D
It's
it's
sometimes
hard
to
get
a
one-size-fits-all
solution.
I'll
pass.
But
my
mistake
in
that,
like
the
whole
idea
around
the
nri
was
that
the
alignment
choices
made
were
actually
opaque.
The
duster
itself.
M
Yeah,
yes-
and
I
think
I
think,
don
brought
that
up
in
terms
of
making
cluster
aware
scheduling
and
in
the
whole
concept
of
like
the
scheduler
assigning
a
pod
to
a
node
and
then
that
node
going
nope.
I
can't
run
it
because
of
opaque
reasons
and
the
scheduler
going.
Okay.
Does
anybody
have
non-opaque
reasons?
They
can't
run.
D
B
D
B
So
so
so,
yes,
that's
the
other
concept
we
had
in
the
past
and
but
also
there's
the
demanding
people
want
to
address
this
problem.
So
as-
and
I
think
about
the
signal
that
might
be
it's
good-
the
foreign
to
discuss
that
one,
but
sometimes
maybe
they
want
more
attention
like
the
numeral
wire
stuff,
I
think
the
when
you
was
gone
or
maybe
you
are
there,
and
people
still
want
that
one.
So
I
said
okay
in
our
cases,
maybe
we
need
to
discuss.
I
totally
understand
your
capacity,
how
how
much
we
can
handle.
B
D
Actually,
what
we
can
manage
as
a
as
a
community-
I
guess-
and
so
we
have
divergent
interests
often
coming
forward,
and
I
just
want
to
make
sure
that
I
understood
that
that
was
a
gap.
N
That
I
mean
that's
a
that's
a
valid
thing,
that
with
the
nri
there's
divergent
right
requirements,
but
just
because
that
there's
lower
you
know
config
doesn't
mean
we
can't
expose
the
config.
You
know
upwards
through
cri,
for
example.
Today
with
cni
we
do
provide.
You
know
all
the
gross
details
of
how
cni
is
configured.
You
know
for
your
pod.
If
you
request
it
cool
all
right,
well,
micah.
D
I
appreciate
that
feedback.
All
I
had
seen
was
a
ten
slide
deck,
so
I
was
actually
really
looking
forward.
O
Lot
of
work
left
to
be
done,
so
I
would
say
it's
like
just
discussion
opener,
because
there
are
a
lot
of
things
to
be
discussed
and
decide
how
to
do
it
in
the
best
way.
Yeah.
B
So
so
so
so
barack
and
I
is
going
to
discuss
tomorrow
and
because
just
like
derek-
and
I
are
both
also-
we
don't
know
like
the-
I
think-
just
not
just
two
of
us,
like
the-
I
think
it's
just
signaled-
also
it's
the
signal
as
the
community
and
there's
the
balance
right
so
ben
both
of
us
desperately
want
helpful.
B
Everyone
can
move
forward
and
we
are
not
the
blocker
for
the
new
idea
and
the
feature
all
those
kind
of
things,
but
on
other
hand
we
also
kind
of
want
to
not
to
make
the
allow
certain
feature
go
out
of
balance
and
the
these
stableness
are
our
quality
and
the
stableness
of
because
ever
a
lot
of
feature
maybe
have
the
particularly
vendor's
interest.
So
we
try
to
be
balanced
here,
so
we
try
to
do
our
best,
but
sometimes
we
may
not
see
other
things.
So
so
we
are.
We
are
we.
B
We
have
a
meeting
tomorrow
and
we
are
looking
forward
and
look
and
discuss
this
in
detail
and
just
like
the
director
just
mentioned
that,
like
the
back
challenge
and
the
pr
challenge
meeting,
that's
really
want
to.
B
We
don't
want
to
do
our
best
to
move
faster
and
but
on
other
hand,
we
also
have
concern
like
the
signal
as
the
community,
because
if
you
notice
that
the
majority
feature
will
came
to
the
signal
and
the
talk
and
and
a
lot
of
feature
used
to
be
like
we
talked
about
the
scheduling,
I
need
to
do
work
but
then
basically,
scheduler
still
push
backs.
Oh
link
to
the
signal
to
handle.
B
It
is
easy
to
put
on
the
node
and
bypass
all
those
controller,
all
those
kind
of
things,
but
actually
there's
also
have
the
risk
and
the
damage
to
the
overall
community
for
overall
project.
So
we
will
please
give
our
next
share
your
opinion
and
brainstorming,
and
we
also
do
our
best
to
help
here
yeah.
B
So
so
we
we
will
tomorrow-
and
I
will
I
we
heard-
there's
the
demand
to
continue
on
some
resource
management,
but
the
resource
management's,
huge
topic
so
give
both
of
us
a
little
bit
time
and
we
figure
out
the
best
plan
and
I
mean
not
not
perfect
plan,
but
we
will
do
our
best
yeah.
D
Thanks
don
okay,
cool.
D
I
guess
just
one
of
the
topics
from
today
we'll
we'll
get
something
sent
out
to
the
mailing
list
on
triage
times
and
thanks
for
joining.
J
Thank
you.
Thank
you.
Everyone
head
down
now
I
I
added
sorry,
I
added
last
item
in
there
just
a
quick
fy,
and
so
last
time
we
talked
last
week
we
talked
there
are
multiple
streams
of
interest
to
continue
the
signal,
the
test,
ci
sub
project
or
group
effort
right.
So
I
worked
with
victor
and
the
joy
and
a
few
others
in
the
community,
and
we
scheduled
a
initial
meeting
this
friday
from
9
to
10
a.m.
Pst
time.
J
B
Thank
you.
Thank
you
for
lunch
here,
yeah
yeah,
I
think,
there's
the
also
on
the
slide
and
alex
also
share
something
about
the
ci
change,
and
I
will
put
into
our
signal
meeting
notes,
and
so
people
can
easily
access
that
one
and
to
take
a
look,
the
new
policy
improve
of
the
kubernetes
ci,
so
I
just
I
will
put
it
there
awesome.
Thank
you
yeah.
Thank
you.