►
Description
Meeting agenda: https://docs.google.com/document/d/1aPgGRl4WewM3txrCYvkepsxLUvGdMG1EzlVfCNeV74M/edit#bookmark=id.ckyeuzl23aar
A
A
Welcome
to
the
August,
9th
2023,
API
server,
Network
proxy
working
group
of
the
Sig
cloud
provider
meeting.
As
always,
this
is
a
cncf
meeting
and
we
are
beholden
to
the
cncf
rules,
which
mostly,
is
please
be
considerate,
inclusive
of
all
your
fellow
members
and
with
that
I'm
going
to
turn
it
over
to
Joseph.
B
Yeah
I
think
under
the
agenda,
the
I
see
three
topics
and
the
first
is
the
agent
Readiness
PR
from
author
Imran
who's
here,
so
Imran,
actually
right
before
this
meeting
I
skimmed,
the
latest
Force
push
and
it
addressed
the
final
outstanding
comment
that
I
was
waiting
on
so
I'm,
ready
to
re-add
the
lgtm
and
approve
and
I
know
I
put
a
hold
on
it,
and
it
was
just
out
of
sort
of
etiquette
to
allow
I
think
mainly
Walter
a
bit
of
time
to
look
as
well,
but
I
think
I.
C
B
Okay,
next
topic
up
is
the
branching
strategy.
B
I
know
that
in
the
last
bi-weekly
meeting
of
this
agenda,
I
had
signed
up
for
this
AI,
but
just
a
day
or
two,
a
guy
opened
I
did
finally
open
that
issue
I
linked
to
it
from
the
meeting
notes,
it's
kind
of
simple:
it's
not
much.
It's
just
what
we
discussed
last
time,
which
is
hey.
Let's
follow
the
sort
of
client
go
pattern
of
the
same
sort
of
minor
version,
Cadence.
A
So,
just
to
be
clear,
if
folks
aren't
familiar
with
the
claim,
go
one
of
the
things
the
client
go
does
is
rather
than
start
from,
I
mean
they.
They
made
the
same
transition
we're
discussing
here,
but
rather
than
just
being
incremental
and
starting
with
one,
they
actually
make
their
minor
version
match
the
kubernetes
minor
version,
so
that
it
is
a
lot
easier
to
understand
the
correspondence
between
the
client
which
client
go
you're
building
and
which
kubernetes
version
that
it
it
sort
of
intended
to
work
with.
So
the
idea
would
be
as
an
example.
B
Yeah
so
open
that
issue,
since
it's
a
big
process,
change
I
think
we
wanted
to
let
a
fair
amount
of
time
go
by
for
people
to
give
comments
and
feedback
before
we
go
and
implement
the
new
either
tagging
or
branching.
So
if
you
have
any
thoughts
or
opinions,
please
go
leave
them
on
that
issue
and.
A
Just
to
be
clear,
one
of
the
kubernetes
contributors,
Jordan
Liggett,
has
an
open
work
in
progress,
PR
to
update
our
go
mod
and
go
some
and
dependencies
primarily
for
grpc,
but
I
believe
this
has
to
do
with
a
CBE.
A
The
top
level
go
mod
goes,
some
are
not
an
issue,
but
the
ones
on
their
connectivity
client
are
because
they
inferentially
go
into
the
cube
API
server
and
we've
had
issues
before
where
we
inferentially
upgrade
the
packages
that
the
cube,
API
server
uses
and,
in
fact,
the
version
of
go
that
it
it's
supposed
to
be
compiled
with,
and
that's
not
how
we
want
to
control
that
in
the
in
cncf
in
the
kubernetes
project.
A
So
if
we
want
to
go
all
the
way
up
to
what
128
does
then,
obviously
we
would
like
to
make
those
changes
in
the
branch
that
corresponds
to
the
kubernetes
128
release.
A
And
in
fact,
what
is
the
main
branch
I
think
is
a
great
discussion
we
should
have,
but
yeah
I
mean
so
right
now
the
main
branch
is
feeding
into
older,
anything
I
think
over
120,
5
or
126..
So
essentially,
what
we're
saying
is
that
we,
we
might
actually
just
make
a
branch
that
corresponds
I
I.
We
should
make
sure
that
we
get
all
the
details
straight
and
I'm
kind
of
hand,
waving
and
I
apologize
for
that.
But
the
gist
is
that
we'd
have
one
branch
for
128
which
would
have
0.28
Branch
tags.
A
It
might
be
the
main
line
for
now
and
then
we'd
have
a
a
slightly
older
branch
which
would
correspond
to
whatever
the
minimum.
We
support
on
the
current
main
line
up
to
127.
A
A
B
Yeah
that
that
sounds
attractive
just
to
reduce,
toil
and
have
consistency.
A
And
it
may
be
worth
checking
on
the
staging
repos
under
kubernetes
are
strange.
Some
of
them
are
replicas
going
out
of
staging,
and
some
of
them
are
replicas
going
into
staging.
That
whole
thing
is
a
bit
of
a
mess,
but
yeah.
C
Should
I
ask
I
mean
my
question
would
be
to
have
all
these
questions
and
discussions
somewhere
they
sent
down
I
believe
we
do
have
an
issue
about
it
right.
If
this
is
captured
as
comments,
perhaps
it
can
later
flow
into
the
readme
or
any
further
documentation,
or
you
really
can
just
say
it
as
a
an
issue
which
can
be
related
on
by
others.
B
Yeah,
that's
fair
and
I
I
noticed
we
haven't
been
taking
notes
so
far.
I
can
try
to
backfill
the
notes
that
we
just
discussed
either
in
this
meeting
notes
document
or
in
that
issue.
B
Ben's
got
a
useful
link.
There.
A
And
is
voice
working,
it
is
now
okay,
yeah,
so
I
mean
we
do
have.
Oh,
we
do
have
the
repo
branching
proposal
that
Joseph
did
I
agree.
It
would
be
nice
to
get
this
all
captured
in
one
place.
I'm,
not
sure.
If
folks
are
okay
using
this
issue
to
do
the
discussion.
A
C
C
A
So
hopefully,
the
the
issue
I
think
it's
issue.
510
is
being
displayed
and
I'm
I'm,
just
suggesting
that
this
is
the
one
that
Joseph
opened
up
and
if
everyone's
okay
with
it,
why
don't
I
I,
like
your
suggestion,
like
I,
think
we
we
need
to
capture
it
all
in
both
the
readme
and
preferably
an
issue
and
I'm,
suggesting
that
for
now,
this
issue
seems
like
a
good
place
to
capture
the
discussion,
and
then
we
should
capture
the
decision
in
the
readme.
C
Right
right,
so,
once
we
finalize
like
what
strategy
do
we
want
to,
we
can
sort
of
create
a
PR
and
capture
everything
that
is
finalized
in
the
the
real
me,
but
up
until
then,
all
our
discussions
questions
just
keep
flowing
into
this
issue.
A
B
Did
you
want
to
did
you?
Do
you
have
some
idea
of
liggett's
urgency
for
his
PR
that
could
block
on.
A
This
so
I
suggested
breaking
it
into
two
pieces
so
that
he
could
get
the
actual
agents,
because
the
the
thought
process
I'd
had
was
that
if
it
was
a
really
urgent
CBE,
the
binaries
would
all
be
at
the
right
version.
If
we
just
broke
it
into
two
commits-
and
he
said
he
didn't
really
think
it
was
that
urgent,
so
I
think
we
have
a
couple
of
weeks.
B
Okay,
next
up
on
the
agenda,
is
cubecon.
B
So,
let's
see
what
is
your
name,
it's
Joel
yeah.
So
how
how
familiar
are
you
with
connectivity,
proxy.
E
Not
hugely
to
be
honest,
which
is
why
I
was
a
bit
well
so
Mike
Mike
reached
out
to
me.
He
was
like
Hey
we're
doing
this
talk.
It
might
be
cool
if
you
could
join
in
and
I
was
like.
E
Said
connectivity,
proxy
and
I
was
like
right.
That's
something
I,
don't
really
know
much
about
and
he
was
like.
Yeah
you've
got
a
few
months.
You
could
learn
it
so
I
mean
I've
opened
the
idea,
but
I
think
I
need
to
go
and
learn
about
it
before
I
say
yes
or
not,
but
yeah.
That's
that's
kind
of
why
I
came
so.
A
One
thing
I
will
suggest
you
know:
I
am
happy
to
help
the
open
source.
Community
I
have
a
bit
of
experience.
I
think
there
have
been
two
talks
on
this
subject
so
back
when
I
was
originally
putting
the
whole
concept
together
to
replace
SSH
tunnels.
I
did
a
talk
in
Shanghai,
you're,
welcome
to
ping
me
or
whatever.
I
can
get
you
a
link
to
the
Shanghai
talk.
A
A
I
I
keep
meaning
to
go
back
and
actually
watch
it
because
I'm
fascinated
to
see
see
what
that
how
that
came
out,
but
the
other
thing
is
I'm
really
buzzing
Monday
through
Thursdays,
but
I
generally
have
my
20
for
Google
on
Fridays
to
do
open
source
work.
So
you
know
if
you
would
like
to
spend
like
half
an
hour
asking
questions
or
having
me
do
a
brain
dump.
I
could
probably
book
that
on
a
Friday
for
you.
E
That
would
probably
be
appreciated.
Yeah
no
I
think
Mike
reached
out
because
it's
looking
like
our
team's
going
to
get
budget
approval
and
I
know.
Historically,
the
the
maintaining
track
has
had
some
issues
Staffing
the
talks
because
of
budgets
of
people
at
the
moment,
and
we
know
what
the
climate's
like.
E
So
my
action
item
is
to
go
and
read
the
enhancement
and
try
and
learn
a
little
bit
about
it.
I'll
try
and
find
your
water
and
then,
if
I've
still
got
questions
I'll
reach
out.
B
I
dropped
a
link
to
the
two
recorded
talks
that
I'm
aware
of
the
older
one
from
Walter
and
then
the
one
from
last
year
in
this
meeting,
notes,
yeah
and
then
Joel
and
Walter.
If
you
have
that,
if
you
have
such
a
sort
of
brain
dump,
meeting
on
a
Friday
feel
free
to
add
me,
I've
gotten
a
I've,
already
kind
of
gotten
a
wealth
of
information
from
Walter,
but
I'll
like
I
would
enjoy
participating
just
to
get
some
more
face
time
with
you,
Joel.
E
Yeah
that'll
be
awesome,
I
appreciate
this
is
a
bit
of
a
weird
situation
and
I'll
just
come
in
randomly
be
like
hey
I'm
enjoy
the
talk
but
yeah.
It's
see
how
it
goes.
B
Okay
and
then
Imran
just
seeing
the
agenda,
there's
one
last
item
from
Imran:
oh
adding
you
as
a
community
member
I,
think.
C
So
I
have
opened
a
PR
on
the
kubernetes
for
the
membership
and
and
I
think,
first
time,
I
added
both
of
you,
you
and
I
think
one
of
the
you
mentioned
that
both
people
can
be
in
the
same
company.
So
I've
added
from
someone
from
my
company
from
Microsoft
and
he
has
replied
back
to
sponsor
I'm,
not
sure
what
are
the
next
steps,
who
approves
the
pr
and
yeah.
A
It's
a
great
question:
it's
actually
a
community
question,
rather
than
as
far
as
I
know,
IL,
gtms
and
plus
ones
that
and
then
I'm
guessing
they,
your
your
Microsoft
person
did
as
well.
So
at
this
point
it's
on
the
owners,
which
you
know
I'm
I'm
an
owner
for
this
project,
but
I'm
not
an
owner
for
for
the
community
project.
So
it's
it's
on
the
owners
of
community
project.
A
A
Did
anyone
have
anything
else,
Ben,
I,
I,
I'm,
not
sure
I
I,
remember
seeing
you
here
before
did
do
we
hear
with
any
questions
or
any
concerns.
A
Cool
okay,
yeah
always
good
to
have
API
Machinery.
Here
we
are
the
interesting
intersection
of
API
machinery
and
cloud
provider,
so.
A
A
Okay,
yeah
I
can
certainly
do
that.
So
your
red
hat,
if
you're
Dave,
if
you're
saying
David
I'm
guessing
you're
a
red
hat.
Yes,
yes,
so
could
I
I
realize
not
everyone
at
Red.
Hat
knows
each
other,
but
Joel
could
could
I
leave
it
on
you
when
we
set
this
up
to
invite
Ben
yeah.