►
Description
[SIG-Network] Ingress NGINX Bi-Weekly Meeting for 20220201
A
Okay,
hello,
everyone.
So
today
is
february
1st
2022..
This
is
sig
networking
residential
nexu
project
meeting.
My
name
is
ricardo.
I'm
gonna
be
your
host
today.
This
is
a
kubernetes
community
meeting
and
we
ask
you
to
comply
with
the
conduct
of
conduct,
which
is
basically
be
excellent
with
each
other,
don't
be
a
jerk
with
no
one.
Any
violation
should
be
reported
to
the
sick,
chairs,
sig
network
chairs
or
the
code
of
conduct
committee
of
kubernetes
and
or
you
can.
A
A
B
I
am
alex
there
a
first
and
yeah,
I'm
just
stopping
by
I've,
been
using
ingress
for
a
long
time,
ingress,
nginx
and
I'm
hoping
to
contribute
to
static,
manifest
fixes
for
different
kubernetes
versions,
because
currently
they
all
target
120
and
I'd
like
to
unblock
it.
A
Nope
all
right
so
yeah
so
alastair.
So
first
of
all
I
have
been
taking
some
sabbatical
time
the
latest
few
weeks.
So
I'm
really
sorry
about
the
the
stock
prs
on
my
name,
but
I
intend
to
unstack
them
this
weekend.
Probably
so
next
thing
on
the
meeting.
Let's
take
a
look
into
the
issue
triage
and
see
what
we
have.
So
I
know
you
added
some
some
pull
requests
here.
C
No
lately,
I
have
been
only
looking
at
pull
requests
so,
as
far
as
I
can
say,
there's
nothing
super
important
regarding
the
issues,
but
we
can.
Of
course
we
can
have
a
look.
A
All
right
so
as
I
am
as
I
I
was
away
on
the
last
meeting,
but
I
watched
the
recording
and
I
seen
that
we
have
some
some
pull
requests
that
we
need
to
review.
I
want
to
just
take
a
jump
into
the
pull
requests
and
I
don't
have
any
subject
from
my
side
that
I
need
to
bring
to
the
meeting.
I'm
still
stuck
on
that
control,
plane
data
plane,
split,
which
I
plan
to
get
back
to
work
so
yeah.
A
A
All
right
so
canary
consistence,
considering
reverse
okay.
Oh
this
is
like
almost
one
year
old,
yeah.
C
Yeah
that
that's
why
I
bring
them
on
the
table.
It's
nothing
really
big,
but
it's
super
old
and
I
think
it's
now
ready
to
actually
be
reviewed.
There
have
been
some
problems
with
rebasing
and
so
on.
Okay,
but
you
already
tagged
it
for
my
for
the
milestone
one
one
two,
so
it
just.
A
I
will
take
a
look
honestly.
I've
been
looking
for
folks
that
know
lua
better
than
I
do,
because
every
time
I
need
to
touch
on
the
lua
code,
I
am
a
bit
afraid
of,
but
I'm
assigning
that
one
to
myself
and
I
will
take
a
look
into
all
of
the
canary
things,
because
this
one
is
really
old.
So
at
least
I
want
to
give
to
the
to
the
contributor
an
answer
if
it's
yes
or
no,
and
why
not
so,
instead
of
just
getting
this
tail,
let
me
let
me
take
a
look
into
that.
Yeah.
A
A
Otherwise
you
are
just
giving
you
are.
You
are
assuming
the
risk
of
me
reviewing
your
code
and
accepting
unity.
Okay,
so
let
let's
move
to
the
next
one,
which
is
this
one
fix,
buggy,
retry
logic
in
ingress,
okay,
this
one!
I
remember
this
one.
I
guess
I
I
did
some
discussion
on
this
and
it
needs
rebase
again,
I'm
sorry
about
it
yeah.
I
I
think
I
think.
C
The
one
that
opened
it
isn't
working
on
it
anymore.
Maybe
we
can
ping
him.
It
needs
to
replace
eight
months
later.
A
Old
yeah,
yeah
and-
and
I
think
I
think
my
my
my
fault
on
this-
I'm
I'm
trying
to
yeah
yeah.
No,
but
that's
that's
that's
not
like
I
I
would
get.
I
usually
get
annoyed
when
I,
when
I
open
a
pr-
and
I
wait
like
six
months
until
someone
says
to
me-
hey
rebase
this
thing
please
so
I'm
I
just
need
some
help
to
review
those
things
as
well,
but
yeah.
Let's,
let's
do
this,
so
I
think
this
is
what.
C
A
A
C
C
A
C
C
B
A
Some
of
those
things
here
but
yeah
this
one
is
with
me.
So
let's
do
this
thing,
so
let
me
add
a
milestone
here.
So
at
least
I
I
may
know
what
I
am
postponing.
Okay,
all
right.
Let's
move
to
the
next
one
support
block
list
yeah.
I
remember
this
one,
so
I
want
to
discuss
this
with
you
as
we
are
here
so
and
I
want
everyone's
opinion.
So,
first
of
all,
let's
forget
a
bit
about
the
wording
here.
If
this
is
like
white
list,
black
list
allow
this.
A
So
in
my
in
this
case,
I
think
that
we
should
just
move
to
block
list
and
just
disregard
the
old
wording,
and
then
we
we
should.
We
should
take
care
of
the
wording
like
a
low
list
or
a
white
list
in
a
future
pr,
maybe
adding
both
of
the
annotations
but
just
migrating
when
someone
adds
like
a
low
list,
this
migrates
internally
to
the
structure
of
whitelist
right.
So
let's
forget
this
discussion
that
we
had
here
for
a
bit.
A
So
the
feature
here
is
asking
for
something
called
block
list
and
they
and
talkers
adds
some
real
user
stories.
A
real
actually
good
case
here,
which
is
I
wanna,
deny
a
specific.
A
I
wanna
deny
a
specific
ip
address
from
of
reaching
my
my
inginx
server,
but
I
want
to
allow
everything
else
right,
I'm
not
sure
from
my
from
my
side
to
be
honest,
in
which
case
we
want
to
do
that
that
we
shouldn't
be
doing
actually
with
network
policies
or
that
we
shouldn't
be
doing
with
firewall
rules
in
front
of
ingress
in
ginx
right.
So
I
imagine
that,
in
like
specific
environment,
develop
development,
development
environments-
like
I'm
running
this
on
kind,
I'm
running
this
on
mini
cube.
I'm
running
this
on
cube
adm.
A
On
the
other
hand,
it's
a
cool
feature
feature
to
have
if
you
are
like
under
attack-
and
you
want
to
block
like
range
of
ip
addresses,
but
you
don't
wanna,
you
don't
want
to
put
your
cersei
down
just
remembering
that,
in
case
of
you
are
like
under
heavy
attack,
and
you
are
allowing
your
inginx
to
deal
with
that
connection,
instead
of
blocking,
with
with
the
load
balancer
in
front
of
that
or
with
the
firewall
in
front
of
that,
you
are
expanding
cpu
and
memory
cycles,
but
now
thinking
on
that,
it's
just
up
to
the
user
if
he
wants
to
use,
if,
if
the
user
wants
to
block
the
the
connections
in
the
the
reverse
proxy,
instead
of
block
the
connections
in
the
firewall
or
whatever
right
so,
let's
say
I
wanna,
listen
from
you
folks,
if
the
complexity.
A
To
be
honest,
I
remember
that
this
didn't
added
too
much.
Complexity
on
on
the
code
was
mostly
was
mostly
ordering
and
some
some
stuff
related,
I'm
just
afraid
of
after
all,
of
the
cvs
that
we
got
and
we
are
still
getting
some
of
them.
A
So
you
are
probably
listening
about
the
new
cvs
in
a
few
weeks,
I'm
sorry,
but
after
getting
those
cvs,
if
we
want
to
add
something
else
that
that
changes,
the
security
layer
right
now,
because
thinking
on
that
like
this
is
our
security
layer
and
we
may
have
situations
where
the
admin
they
add
allow
lists
or
block
lists
because
they
have
like
a
protected
web
services
running
behind
english
and
china
x.
That
should
be
reached
only
by
by
a
range
of
ip
addresses.
A
So
that's
it.
I
I
just
just
want
to
know
your
opinion
on
that
folks.
So
what
do
you
think
about
that?
If
you
are
fine
on
that,
if
someone
can
help
me
reveal
that
thing,
if
is
not
able
to
finish
that,
we
can
just
ask
him
hey
so
curious.
We
want
to
close
and
reopen
this
on
another
pr,
but
anyway
we
need
more
people
more
eyes
and
more
thinking
on
security
perspective
of
things
right.
A
So
that's
not
just
accepting
the
future
the
future,
because
okay,
that's
a
cool
feature,
but
what
are
the
security
implications
if
we
miss
something
here
if
we
left
something
past
so
this
has
been
mostly
my
concerns
on
the
last
few
days
last
few
weeks,
and
I
want
your
help
on
that.
Okay,
on
this
discussion,
so
thoughts
on
that
someone
will
get
assigned
to
this
pr
as
well.
C
C
But
on
the
other
hand,
I'm
also
with
you
regarding
the
the
security
issues
related
to
the
cpes
that
we
might
have
the
next
time.
So
maybe
we
should
just
postpone
this
one
or
assign
a
lower
priority
to
yeah,
to
put
it
back
on
the
queue
again.
B
Yeah,
I
mean
it's
a
feature
of
nginx
plus
right.
I
think
the
the
actual
output
change
is
very
small.
Looking
at
the
template
nginx.tmpl,
it's
like
deny
one
thing
allow
all,
but
I
totally
understand
it's
like
this
is
the
code
that
you
don't
want
to
break
right
and
if
you
have
an
author
and
then
like,
I
also
didn't
really
see
any.
I
mean
I
was
given
the
thread,
but
I
didn't
see
any
use
cases
for
it.
B
This
is
something
that
only
the
developer
level
would
want
to
do,
and-
and
I
can
see
that
as
a
developer
user,
who
only
has
access
to
create
ingress
resources
in
the
cluster.
This
can
be
handy.
But
that
is
exactly
where
the
can
of
worms
for
cvs
happens.
We
have
a
shared
ingress
resource
and
a
developer
is
only
allowed
to
create
ingresses
but
modify
the
whole
controller.
A
Yeah,
so
I
I'm
gonna,
I'm
gonna,
my
my
thinking
on
that
is
that,
like
we
don't
have
any
issue
that
requests
this
feature,
it's
just
something
that
talkers
tackle.
A
I
am
thinking
as
we
say
that
we
we
should
not
be
afraid
of
new
features
because
of
just
because
of
security,
but
we
need
to
be
careful
right.
So
my
thought
on
that
right
now
and
something
that
I've
been
talking
with
jamie
is
that
I
wanna
before
I
start
accepting
new
new
huge
features.
A
A
Some
some
are
security
before
just
jumping
into
new
features
right,
so
I
won't
close
this
thing.
No,
I
think
that
one
thing
that
we
can
ask
and
talk
with
talkers
later,
is
that
hey
we
we
we
had
a
chat
in
the
ingress
in
chinax
development
and
feel
free
to
join
us.
If,
if
you
want,
I
know
that
talkers
probably
is
on
is
on.
China
is
on
china
time
zone,
so
we
still
owe
for
for
apec
folks,
better
a
better
meeting
time
as
well.
A
Anyway,
we
can
reach
him
and
and
say,
hey
talk
here,
so
we
discuss
it.
We
don't
think
that
right
now
is
the
right
time
to
get
this
thing
just
before
we
can
fix
this
other
security
things,
but
still
that's
we.
We
are
not
denying
this
feature.
We
are
just
like
been
stepping
on
eggs
right
now,
right
so
yeah
trying
to
be
really
careful
with
that.
Okay,
then
let
me
just.
C
So
I
also
found
the
issue
the
original
issue
that
it
it
was.
The
request
was
referring
to
I'm
just
doing
yeah.
Do
I.
A
C
A
Proxy
protocol,
I
think
so
I
never.
D
Okay,
so
if
the
allow
denials
are
using
the
nginx
access
module,
it
will
always
only
look
at
the
actual
source
and
destination
address.
It
does
not
have
a
feature
to
the
look
at
the
proxy
protocol
address,
so
this
will
be
a
caveat
if
you
let
that
pr
through.
A
A
I
remember
having
an
issue
on
past
still
when
I
was
talking
about
that
with
alejandro
that
when
you
enable
the
proxy
protocol,
it
takes
the
x
forwarded
for
header
or
something
like
that
as
the
original
yp
as
the
original
ap
yeah.
D
A
Can
can
you
can
you
can
you
so
I
can
see
here
like
that.
We
have
the
real
ip
header
and
the
proxy
protocol,
but
can
you
can
you
can
you?
Can
you
just
confirm
that
and
maybe
comment
on
that
issue
as
well
or
on
the
request
yeah
yeah?
So
it
would
be
really
helpful
if
you
want
to
also
take
a
look
into
that
review
and
say
hey,
this
is
okay.
This
is
not
okay.
We
are
missing
something
I
am
mostly.
A
If
you
tell
me,
okay,
ricardo,
this
is
fine
like
I
took
a
quick
pass
and
it's
fine.
I
just
need
someone
else,
but
get
me
to
take
a
look
and
accept
things,
because
if,
if
it's
just
me
I
would
say
yeah
it's
fine
and
we
can
have
another
cd
so
feel
free
to
take
a
look
into
that.
Okay,
I
cannot
find
that
okay.
D
Sounds
good
sounds
good
because
yeah,
that's
something
I
have
personally
experienced
it's
a
missing
gap
in
the
nginx
access
module.
D
A
Yeah,
you
can
come
later
and
ask
me
some
for
some
membership.
Okay
sure
should
we
probably
should
be
better
okay,
this
one,
including
lua
libraries
yeah.
I
remember
this
one.
I
don't
wanna
right
now
touch
this
one,
I'm
really
tempted
on
closing
this
one.
I
think
that
I
think
that
right
now
this
doesn't
add
any.
A
It
doesn't
add
any
any
new
thing
it
doesn't.
It
doesn't
solves
any
any
problem
other
than
just
having
a
jwt
on
on
as
part
of
the
build
right.
So
if
we
got
something
else,
saying
hey,
we
have
this
new
annotation
or
we
have
this
new
feature.
We
are
allowing
ingress
in
gynex
to
parse
jwt
headers,
and
we
have
like
the
secret
key
this
shared
key
or
this
certificate
that
I
can
open
and
close
and
direct
to
some
other
place.
A
I
think
it
would
be
fine,
but
right
now
and
after
that
card
problem
that
we've
had
on
past,
I
don't
want
to
add
any
any
new
thing
in
in
lua
data
path
that
may
that
may
cause
a
recorded
import
that
may
cause
just
just
something
more
complicated
and
without
any
gain.
Okay,
so
I
I
will
take
a
look
into
that
noah
and
I
will
answer
them
and
say:
yeah.
B
C
That's
fair,
just
just
just
write
a
comment
and
then
it
should
be
fine.
I
think.
B
A
Yeah,
so
so
yeah
in
this
case,
specifically,
I
I
mean
like
the
idea
of
using
jw3.
It's
fine,
I'm
just
afraid
of
turning
in
gynex
much
more
an
application
server
than
just
a
reverse
proxy
right.
So
I
will
I
will.
I
will
take
a
look
into
that
and
I
I
will
write
some
justification
and
I
will
talk
with
james
and
maybe
being
you
all
and
say,
hey
we.
This
is
recorded
and
if,
if
you
don't
agree
with
that,
let's
just
discuss
some
slack
on
other
places.
A
Yeah
there
are
some
there
are
other
solutions
and
also,
as
I
said,
like
I'm
planning
doing
some
more
exploitation
on
on
in
js
on
ninja
next
javascript
that
you
can
run
javascript
inside
and
in
that
case
it's
supported
right
because
it's
a
library,
that's
already,
that's
already
supported
by
nginx,
it's
part
of
the
core,
so
it
would
be
something
like
okay.
If
you
want
to
deal
with
that,
you
can
just
enable
that
thing
in
your.
A
That's
open!
I
will
take
a
look
into
that.
Okay,
this
one
is,
is
yours
right
almost
there?
No,
that's!
Not!
Yours
fix
home
shard,
remove
capabilities,
api
version,
four
customize,
the
parts
file.
Okay.
What
is
this
thing?
A
C
Yeah
so
so
people
seem
to
be
okay
with
it,
but
somewhere
it
got
lost.
C
And
then
I
asked
long
and
how
to
follow
up
on
it
and
only
yeah.
The
only
reply
was
that
we
from
long
that
that
we
should
create
an
issue
to
summarize
the
problem.
B
B
So
if
you're
doing
it
with
customize
or
kubecuttle
template
capabilities,
api
versions
is
always
going
to
be
empty
and
you'll
never
get
this.
No
matter
what
you
said,
it's
a
problem
and
it's
probably
going
to
affect
the
static
manifest
and
when
you
merge
this,
I'm
probably
going
to
have
to
rebase
all
mine
again
too.
E
A
C
A
B
You
can
actually
you
can,
if
you
do,
minus
minus
validate,
which
is
a
nice
little
workaround
to
make
helm
template
connect
to
the
api
server.
But
it's
silly
right.
If
you
have
a
service
service
monitor
enabled,
then
you
want
the
service
monitor.
A
B
A
B
E
B
A
No,
no
problem:
okay,
let's
move
to
the
next
one,
the
last
one
so
canary
by
query.
Do
we
want
to
support
something
like
that?
So
this
is
something
that
I
wanted
to
bring
as
a
discussion
as
well.
I
keep
seeing
prs
of
support
canary
by
query.
Support.
A
Can
every
by
country
support
canary
by
the
moon
phase
if
it's
just
a
full
moon
or
if
just
like
a
starting
moon
or
if
it's
winter
in
the
north-
and
I
mean
just
kidding,
but
but
I
I
don't
know
how
sustainable
is
this
is,
is
like
we,
we
we
accepting
canary
by
any
of
those
situations
right-
and
I
don't
know
how
much
this
makes
sense-
to
keep
adding
canary
to
keep
adding
canary
support
for
yeah.
Whatever
that's.
B
A
A
I
would
just
say
we
need
to
stop
accepting
like
canary
by
energy
right,
so
we
should
probably
define
what
we
do
accept
as
a
canary
and
what
we
are
not
going
to
accept
as
a
canary
later.
So
as
far
as
I
remember,
we
accept,
like
the
header,
one,
the
cookie
one,
then
the
query
one,
the
the
weight
one
and
there
is
another
one
with
like
the
joey
p,
with
joey
p
thing,
I
guess
like
country,
state,
etc.
A
So
I
will
review
this
one.
I
will
add
a
milestone
on
this
one,
but
I
want
to
just
make
a
contract
here
that
maybe
we
should
stop
getting
a
pr
sweet.
Let
me
take
a
look.
I
guess
there
were
like
canary
canary
consistency
this
one
and
if
you
take
a
look
at
the
other
one,
we
got
like
carry
canary
backing
name
then,
and
this
is
the
the
metrics
blah
blah
blah.
Well,
I
remember
seeing
some.
B
E
A
B
A
B
A
A
So
if
this
makes
sense,
I
will
take
a
look
into
that
with
some
with
some
some
some
some
fine-
and
I
just
wanna-
be
careful
with
other
canaries
demands
as
well
in
the
future,
like
hey,
I
wanna,
add
canary
by
whatever,
let's
discuss
before,
just
jumping
and
accepting,
because
every
new
thing
that
we
add
actually
adds
also
a
complexity
on
the
code
that
I'm
not
sure
I
want
to
keep
on
on
the
lua
code.
Okay,
that's,
okay!
This
one
is
is
with
me
as
well
all
right,
so
that's
35!
A
We
have
25
more
minutes.
So
do
we
do
we
have
any
any
any
item
we
want
to
discuss
outside
of
the
agenda
that
you
folks
forgot
to
add
here
some
stock
pr
something
that
needs
my
attention
noah's
attention.
Other
folks,
attention.
B
A
A
One
second
make
a
ghost
yeah:
you
should
be
able
to
share
your
screen.
B
Fine
yeah
there's
nothing
in
here
cool,
so
the
basic
idea
was
to
take
all
of
the
bash
code.
The
here
docs,
which
were
values.yaml,
make
them
into
separate
files,
then
run
home
template
over
those
and
then
type
those
into
customize.
B
What
that
caused
was
that
they
were
all
reordered
and
the
reason
to
do
that
was
because
the
original
helm
template
with
no
value
specified
targets,
kubernetes
1.20
only
which
means
all
of
those
pr's
or
sorry,
all
of
those
lines,
capabilities.coup
version
or
whatever
don't
work
you
just
get
whatever
the
helm
default
version
is
which
was
1.20.
B
B
So
this
is
the
new
script
with
no
here
dogs,
and
all
this
pr
does
is
change.
Oh
I'm
on
main
okay,
anyway,
the
new
pr
brings
in
this
line.
So
we
now
generate
for
119,
120,
121
and
122..
B
A
Cool,
so
right
now
we
are
just
stuck
on
the
generation
of
the
new
manifests
right
and
that's
that
pr,
you,
like
your
older
pr,
got
accepted
the
8000.
I
guess-
and
this
is
just
the
generation
one
right
or
exactly.
B
So
all
this
one
does-
and
I
I
don't
know
why
I
didn't
jump
to
it
and
there's
a
lot
of
files
in
here.
Github
doesn't
like
me
all
this
one
does
is
uncomment
that
line
where
it
it
currently
generates
only
for
1.20,
oh
what's
it
called.
A
Okay
sounds
good
yeah,
so,
as
we
just
accepted
that
thing,
probably
you
just
wanna
rerun
this
right
to
rerun
the.
B
Manufacturer,
I
was
not
thinking
correctly.
None
of
the
values.yaml
have
service
monitor,
enabled
because
that
wouldn't
make
sense.
A
Okay,
okay
sounds
good
yeah,
so
I
would
take
a
look
into
that
and
sounds
good
to
me.
So
I
think
that
you
can
also
add
in
your
kubernetes
targets
123
in
that
pr
and
then
we.
A
A
Yeah
yeah
yeah.
We
are
not
right
now
we
are
not
deprecating
the
119.
The
first
thing
that
I
want
to
deprecate
actually
is
the
legacy
branch
right,
so
the
118
the
one
before
119,
but
still
it's
a
hard
discussion
that
we
we
need
to
have
in
probably
in
another
meeting.
I
want
james
on
this
meeting
as
well,
so
we
can
take
the
decision
and
establish
a
timeline
and
start
announcing
to
users
that
we
are
going
to
stop
supporting
old
versions.
A
But
before
that
I
want
to
establish
with
you
folks
that
plan
about
ingress
class,
and
this
is
something
that
I
want
to
bring
to
you
actually.
So
if
we
have
yeah,
we
do
have
some
more
minutes.
So
thank
you.
I
I
would
take
a
look
into
that
alexa,
but
sounds
good
to
me
so
about
english
class
and
as
as
james
comes
back,
I
will
bring
that
discussion
with
him
as
well.
A
But
I've
been
struggling
a
lot
with
ingress
class
issues
and
seeing
users
complaining
about
ingress
class,
and
how
can
we
deprecate
annotation
if
everybody
uses
annotations
and
and
etc,
and
in
fact
they
got
their
point
and
they
are
right
right.
So
this
was
something
that
I
was
talking
with:
rob
scott
and
as
well,
and
what
happened
is
that
we
we
stuck
in
beta
version
in
ingress
beta
for
a
long
time.
A
I
think
was
two
years
and
a
half
almost
three
years,
so
users
they
were
used
to
using
grass
class
in
a
way
even
being
annotation
or
being
ingress
class
by
name
and
not
the
ingress
class.
The
way
that
we
implemented
on
the
version
one
later,
which
takes
the
spec
controller
and
does
every
all
of
that
complexity
that
we
we
saw
right.
A
So
let's
say
let
me
just
okay:
what's
this
one,
it's.
B
The
yeah,
the
controller,
now
watches
all
ingresses,
even
if
it's
not
the
right
class,
which
means
every
time
you
create
an
ingress
have
like
this
is
really
bad
right,
because
we
have
validating
edition
web
hooks.
B
A
Yeah
yeah,
so
it's
it's!
It's
bad!
It's
not
it's!
Not!
It's
not
good!
It's
bad!
I've
been
talking
internally.
I've
seen
actually
some
comments
internally
on
on
the
company
that,
like
this,
our
ingress
does
this
way.
We've
got
a
huge
thread
on
sig
networks
like
channel
about
like
what
we
should
do,
and
why
should
if
we
should
stop
stating
that
annotation
is
deprecated
or
not,
and
this
got
like
direct
impact
in
things
like
start
manager
right
so
site
manager,
folks
they
were,
they
were
actually
asking.
A
So
one
thing
that
I
was
talking
with
with
james
was
we
we've
made
a
mistake
actually
of
blindly
as
accepting
the
idea
that
annotation
was
deprecated
and
just
just
saying
to
folks
hey.
A
If
you
want
to
migrate
to
version
one,
you
should
stop
using
annotation
and
you
just
need
to
start
using
ingress
class
and
even
not
thinking
about
ingress
class
by
name
thing
and
I'm
really
tempted
about
writing
a
document
on
how
we
should
revert
that
right
so
and
bring
that
discussion
to
the
table
that
if
we
should
revert
that
and
how
we
should
revert
that,
then
how
is
this
is
going
to
impact
folks
because
the
first,
I
don't
know
if
you
remember
noah,
but
the
first
big
issue
that
we
got
was
like
ibm
using
that
yeah.
E
A
E
A
That
that
moment,
a
red
flag
should
be
raised
for
us
like
hey,
we
need
to
roll
back
that
thing,
and
I
expect
that
not
to
be
that
late
to
roll
back
or
at
least
to
bring
something
that
allows
users
to
have
the
old
behavior
without
stating
this
is
deprecated
or
that
is
deprecated
and
then
getting
ingress
class
working
properly
on
the
environment,
the
way
they
want.
A
I
know
that
there
are
some
validations
on
the
kubernetes
api
server
side,
which
is
not
controlled
by
us
like
if
you
use
annotation
or
you're
using
grass
class,
you
are
going
to
get
the
denial
from
the
api
server,
but
again
from
our
side
from
the
validation
web
hooks
from
the
controller.
I
think
that
we
can
and
we
should
do
better
right,
so
we
should
have
listened
to
the
users
back
in
that
time.
A
We
should
that
we
should
have
brought
that
discussion
to
the
signature
community
that
time
and
we
have
failed
that
time
and
specifically,
I
failed
that
time.
So
I
wanna
start
writing
something
about
that.
I'm
still
in
the
middle
of
my
sabbatical
stop
at
least
refreshing
my
my
head
a
bit,
but
I
I
want
to
bring
that
and
if
you
wanna,
if
any
of
you
wanna,
take
on
that
and
bring
on
not
a
cap
basis,
because
I
am
like
I,
I
don't
want
to
have
that
cap
processed
inside
ingredients
in
ginx.
A
But
if
any
of
you
want
to
bring
that
as
a
document
or
something
like
that
or
I
can't
just
start
scratching
a
document
on
what
is
the
problem,
and
how
can
we
fix
that?
I'm
just
open
to
that
discussion
and
I
think
that
we,
we
don't
need
to
wait
every
two
weeks
to
discuss
that.
I
just
want
to
make
that
a
synchronous
yeah.
I
agree.
C
I
think
you
just
start
creating
a
mk
docs
or
something
like
that,
something
that
we
can
share
and
where
we
can
also
add
our
opinions,
and
I
also
think
about
that.
Maybe
we
can
have
a
separate
meeting
or
a
discussion
or
whatever.
Besides
the
community
meeting
about
that
topic,.
A
Yeah,
I
agree
with
that.
So,
if,
if
you
so
that's
not
like
a
an
ask,
that's
just
like
a
question.
So
if
you
have
some
time
to
just
scratch,
something
go
ahead.
Otherwise
I
will.
I
will
add
that
to
my
to-do
list
and
I
will
start
just
scratching
something
and
really
I
I
don't
want
like
a
20
pages
document.
I
want
just
like
the
problem,
the
possible
solutions,
and
we
should
probably
discuss
the
design
over
that
and
and
think
on
on,
like
on
the
support
on
the
sre
side.
A
How
is
this
going
to
impact
like
users
from
ingress
and
china
x?
How
is
this
going
to
impact?
Who
is
going
to
implement
that?
If
that's
me
or
if
that's
someone
that
want
to
tackle
on
on
golem
things,
but
I
I
want
to
something
really
simple,
but
yet
like
a
justification
on
why
we
are
rolling
back
and
if
it's
fine
for
the
community
as
well,
because
I
think
that
that
thing
we
should
do
better.
A
D
A
Of
things
and
to
be
honest,
this
is
this
is
confusing
even
for
me
sometimes
when
I
am
looking
into
the
implementation-
and
I
say
wait,
why
am
I
doing
this
and
then
I
just
remember
that,
like
how
things
they
are
tired
right,
so
yeah,
we
need
to
do
better
in
that
case,
okay.
So
no,
if,
if
you,
if
really
just
if
you
have
some
times
to
scratch,.
A
C
I
can
initiate
that
that
discussion,
but
at
some
point
you
guys
will
have
to
comment
because
sure
I
I
don't
have
the
expertise
on
on
that
one.
But
I
can.
I
can
initiate
the
whole
thing
if
you
want
to.
B
A
Best
and
like
the
user
stories-
and
we
can
start
discussing
that
yeah-
it's
fine
for
me,
yeah
sure
the
the
implementation
side
is
just
the
last
thing
we
should
do.
We
should
discuss
how
how
how
users
they
expect
that
to
work.
Is
that,
like
ingress
class
by
name,
if
that's
annotations,
and
even
we
can
just
release
a
version
too
call
it
like.
We
made
a
mistake
and
we
are
sorry
now
you
can
use
ingress
class
on
version
two.
A
The
way
that
you
use
I'm
just
kidding,
but
anyway
yeah
we
will
find
we
will
find
a
way
yeah
yeah,
all
right,
so
we
have
still
ten
minutes.
I
have
just
another
meeting
to
jump
in
ten
minutes.
So,
if
everybody's
fine
with
the
meeting
today,
I
will
end
that
otherwise
anybody
have
anything
else
that
you
want
to
bring
to
discussion.
Folks,
I'm
not
I'm
not
kicking
anyone
out
of
the
meeting.
I
just
say
that
I
need
five
minutes
before
my
meeting,
but
that's
okay.