►
From YouTube: SIG-Kube-Proxy Bi-Weekly Meeting (APAC) for 20220601
Description
SIG-Kube-Proxy Bi-Weekly Meeting (APAC) for 20220601
A
Hi
everyone,
so
this
is
first
june
2022.
We
are
at
the
kpng
apec
meeting.
All
right
so
just
remember
reminder
that
this
is
under
this
cncf
code
of
conduct,
which
is
just
that
be
excellent
to
everyone,
all
right
cool.
So
I
think
sanskar
and
victor
first
time
attendees.
Who
would
you
like
to
introduce
yourselves.
B
A
I
I
recently
started
converting
kubernetes
and
I
was
thinking
how
I
could
start
contributing
to
network
mode
and,
like
I
came
across
the
kpng
project
and
like
q
proxy
in
general,
and
I
thought
that
this
could
be
a
good
place
like
something
interesting
for
me.
To
begin
with,
awesome,
welcome
and
victor.
B
Yeah,
I'm
the
same
way.
I'm
just
learning
and
I've
been
learning
a
lot
about
overall
system,
stuff
networking
and
all
that
I
yeah
try
to
I'm
really
interested
in.
You
know
cscf
stuff,
so
just
trying
to
learn.
A
Cool
welcome
all
right
so,
just
before
I
hit
the
record
button,
shane
asked
me,
but
how
he
can
pick
up
some
good
first
issues
or
bite
size
issues
in
kpng,
so
there
are
like
a
bunch
of
good
first
issues
over
here
like
something
like
this,
which
you
can
pick
up,
which,
which
is
for
like
adding
make
targets
for
grpc,
compilation
and
stuff
like
that.
So
this
will
basically
give
you
an
idea
on
all
the
grpc
bits
for
kpng
and
then
I
think,
there's
ton
of
context
in
this
description
as
well.
A
C
A
All
right
so
so
there
are
things
like
adding
unit
tests,
which
would
also,
I
I'm
not
sure
if
that
can
be
done
in
the
afternoon,
but
it
would
definitely
give
you
some
idea
around
how
that
back
end
is
written,
written
and
walk
you
through
the
code
base
for
sure,
and
then
you
can
cover
that
code
based
in
unit
tests,
like
you
know,
things
like
ipvs
ip
tables.
C
A
All
right
so
some
of
the
main
areas
where
we
need
help
like
the
cap,
so
we
want
to
make
progress
on
the
kept
aspect.
A
A
If
you
go
to
kpng
back
ends,
there
are
a
ton
of
back
ends
over
here
right
and
most
of
these
back
ends
are
running
in
in
the
ci
and
not
all
conformance
tests
are
passing.
So
our
second
area
is
to
make
sure
that
all
of
the
back-ends
pass
the
kubernetes
conformance
test.
A
So
you
would
find
a
ton
of
issues
which
have
things
like
this
right,
which
have
context
on
which
test
fails,
and
things
like
that-
and
you
know
there
are
quite
a
few
issues
over
here
for
ipv,
v4
or
ip
tables,
and
you
know
stuff
like
that.
So
this
is
another
area
wherein
we
want
to
have
some
robust
unit
tests
and
also
make
sure
that
the
back
ends
pass
conformance
test.
A
Apart
from
this,
there
were
a
couple
of
things
added,
like
you
know,
sctp
support,
and
you
know
other
things
like
revamp
like
like
nft
back-end,
comes
with
a
different
type
of
an
approach,
and
maybe
the
other
back-ends
can
use
that
approach.
But
that's
like
optimization-
and
you
know
stuff
like
that,
but
kept
and
making
sure
the
test
pass
is
like,
like
they're
like
the
highest
priority
right
now.
So
if
you
can
provide
some
help
around
that
that'd
be
like
super
cool.
C
D
C
A
A
All
right,
well,
apart
from
this,
the
other
update
that
I
have
is,
we
were
finally
able
to
merge
the
user
space
back
end.
Is
linux
backend
pr
in
cupping,
and
I've
not
edited.
C
A
Yay,
so
I've
not
edited
in
ci
as
of
now,
but
it's
it
it's
working
it.
It
does
work
so
there's
a
follow-up
pr
required
to
edit
in
ci
and
maybe
so
it's
only
the
support
is
added
only
for
ipv4
and
not
ipv6,
and
what
else
there
may
be
some
conformance
test
failings
have
to
create
an
issue
with
which
all
conformance
tests
are
failing
and
all
of
that
stuff.
So
this
is
sort
of
like
a
toy
back
end
that
we
can
use
to
try
out
different
things.
A
So
there's
there's
probably
another
way.
So
I
think
this
issue.
A
Has
that
covered,
wherein
you
know
added
in
ci?
The
other
thing
is:
maybe
we
want
to
rewrite
it
using
the
diff
store
implementation,
so
this
dipstick
implementation
is
how
nft
backend
is
written,
which
is
slightly
different
from
how
back-ends
are
giving
the
original
proxy.
A
A
Cool
yeah
sam,
I
was
just
trying
to
see
what
all
issues
I
can
probably
like
pick
up,
or
you
know
something
that
I
can
get
started
with
so
yeah
that
really
like
what
he
suggested
was
I
I'll
try
to
do
some
of
that
and
then
hopefully
I
can
get
involved
more
and
more
that'll
be
awesome.
Like
you
know,
we
like,
as
of
now
not
many
people
are
working
on
kpng.
So
if
you
can
help
out
with
these
issues
and
the
likes
that
we
create.
A
Another
area
I
see
is
okay,
so
there
are
a
ton
of
things
around
like
specifics
of
these
back
ends
like
iptables
or
ipvs,
but
I
think
I
don't
have
much
expertise
on
these
back
ends.
So
I
I
wouldn't
comment
so
hana
month
and
vivek
have
like
a
lot
of
context
around
these
back
ends,
so
they
would
in
a
better
position
to
talk
about
that.
I
can
talk
about
user
space
back
end
a
bit
to
give
some
context.
A
This
is
just
a
personal
thing
but
like
if,
if
you
find
it
useful
as
well
like
you
know,
go
ahead
and
try
to
dig
it
through.
One
of
the
things
that
I
would
like
to
mention
is
like
for
any
back
end.
A
One
of
the
things.
The
way
you
can
make
sense
out
of
that
back
end
is
through
this
sync
dot,
co
or
anything
of
that
equivalent
wherein
there'll
be
like
a
setup
function
which
would
like
initialize
that
back
end,
and
then
these
methods
sort
of
dictate.
The
terms
right.
Something
like
set
service
which
so
cupping
is,
is
designed
in
a
way
wherein
there
is
a
coupling
server
which
gets
serviced
and
endpoint
resources
from
the
kubernetes
api
server,
and
then
it
triggers
events
on
a
copying
backend.
A
So
whenever
there's
a
new
service
created,
it
would
call
this
set
service
method,
so
copying
server
would
call
this
set
service
method,
which
would
then
eventually
go
and
talk
to
the
backend.
So
here
the
back
end
would
do
something
like
service
update
or
service
ad.
Similarly,
there
is
something
like
delete
service
and
then
set
end
point
and
delete
input,
so
I
won't
go
into
like
you
know
the
code
walkthrough
or
whatever,
but
this
is
usually
a
good
place
to
start
start,
every
backend
has
something
like
set
service.
A
A
To
the
existing
cube
proxy,
even
the
current
cube
proxy
has
stuff
like
this.
If
I
can
recall
correctly
true,
but
the
current
implementation
of
q
proxy
is
entry,
whereas
a
back
end
comes
with
its
own,
go
mod
and
its
own
dependencies
and
the
back
end
can
implement
its
own
stuff
and
it's
not
running
in
one
process.
A
Any
existing
implementation
of
q
proxy
has
everything
running
in
one
process,
so
you
will
have
kubernetes
business
logic,
bits
and
the
networking
bits
in
the
same
process
wherein
this
is
what
cupping
is
trying
to
abstract
at
the
at
a
node
level,
wherein
a
back
end
can
do
its
own
dumb
thing,
that
is,
to
implement
the
proxy
and
not
care
much
about
kubernetes
logic
got
it
yeah,
thanks
cool
and
and
the
interface
for
this
sync
and
all
of
that
stuff
lies
in.
D
A
A
We
talked
about
this
at
rejects
in
spain
with
so
I
think
this
is
d1
and.
A
A
A
This
was
in
2022
as
well,
so
I'll
I'll
post
a
link
to
this
after
after
this
call.
So
so,
if
you,
if
you
want
to
know
about
cupping
101,
there
is
this
video
and
then,
if
you
want
to
know
about
how
to
add
packets
from
user
space
perspective,
then
there
is
another
thing.
So
these
are
like
couple
of
information
bits
that
you
can
consume
to
know
about
camping
and
things
like
that.
So
yeah,
that
was
my
shameless
plug,
all
right
cool.
So
this
is
all
I
had.
Is
there
anything
anyone
else?