►
Description
Kubernetes SIG Network Ingress NGINX Subproject Biweekly Meeting for 20220104
A
Hello,
everyone
today
is
january
4th
2022.
This
is
our
first
meeting
of
ingresson
genetics,
sub
project
for
sig
networking
and,
as
always,
this
is
a
kubernetes
community
meeting
and
we
ask
you
to
comply
with
the
cncf
code
of
conduct,
which
basically
means
be
excellent
with
each
other.
Any
violations
should
be
reported
to
myself
or
to
ricardo.
If
you
have
any
issues,
please
let
us
know
all
right
and
with
that
go
ahead
and
get
started.
So
I
pulled
up
some
of
the
open
items
that
from
last
year
that
we
didn't
get
a
chance
to
look
at.
A
B
B
C
A
D
Yeah,
so
I
I
spoke
with
him.
I
just
asked
him
to
split
this
into
two
pr's
so
because,
actually,
what
he's
doing
is
he's
running,
changing
the
script
and
then
running
that
script
and
showing
what
has
changed
right,
and
I
I
have
asked
him:
hey,
just
just
change
the
deployment
script
and
leave
us
to
create
that
manifest
as
soon
as
we
make
a
new
release,
so
we
can
know
exactly
what
change
it
or
not.
D
That's
that's
fine
to
have
all
of
these
in
that
same
pr,
but
I
I
think
it
would
be
good
to
just
to
change
the
scripts
and
then
and
then
we
can.
We
can
take
a
look
into
that
later.
I
seen
that
he
answered
me
in
slack,
I'm
just
a
bit
busy
this
beginning
of
the
year,
so
I'm
not
that
responsible
in
in
slack,
but
I've
seen.
Actually
he
opened
a
new
pr.
If,
if
you
go
back
to
the
pr's,
you're
gonna
see
another
pr
of
him
just
with
the
manifest
so
yeah.
D
So
I
I
just
need
to
take
a
look
into
that,
but
I
I
I'm
really.
Okay
with
that,
I
think
that
the
approach
was
really
cool,
using
helm
and
customize,
etc.
I'm
just
worried
about
that
that
we
missed
something
because
of
the
the
size
of
the
pr
like,
as
we
are
mixing
the
proper
script
with
the
result
of
the
script
I'm
afraid
of
like
this
is
an
extra
extra
large
and
we
we
are
just
missing
something
important
to
be
reviewed
because
of
the
generated
file.
E
D
D
A
D
Make
sure
to
I
wanna,
I
wanna
finish
those
by
the
end
of
this
week,
because
we
have
another
bug
fix
that
someone
found
and
I
sent
the
patch
and
then
they
opened
that
there
is
another
pr.
So
I
I
still
think
I
still
folks
for
a
new
release.
We
have
a
bunch
of
bug
fixes
between
the
last
three
days
and
this
new
one
of
a
lot
of
people
jumping
in
and
fixing
okay,
I
just
yeah.
It
wasn't
just
a
busy
end
of
here
anyway,
but
yeah.
We
we.
A
D
Yeah,
okay,
all
right
I
mean
I
mean
we
shouldn't
we
shouldn't
we,
we
shouldn't
update
the
libraries
for
123.,
because
then
the
library
is
updated,
203
they
remove
all
of
the
they
remove
a
lot
of
stuff
right
or
was
122..
I
can't
remember
anyway,
I'm
I'm
I'm
sticking
with
the
old
library
versions,
because
I
am
I'm
afraid.
A
A
Awesome:
okay:
that
was
the
big
one
that
I
know
that
people
have
reached
out
for
that
we're
asking
about.
Let
me
go
ahead
and
pull
up
the
new
issues
just
start
from
the
top.
I
guess
this
one
was
open
very
recently.
I
think-
and
I
was
gonna
again
I'm
still-
and
I
probably
should
get
better
at
this.
So
really,
as
I
say
this
I'm
convincing
myself,
I
should
probably
assign
this
to
myself
the
the
the
clarification
between
in
glass,
ingress
class
name
change
with
the
backwards
compatibility.
A
D
Okay,
first
of
all
this,
this
is
not
something
that
we
control.
This
is
something
that's
coming
from
kubernetes
yeah
yeah
right.
So
so
I'm
not
gonna
say
that
I
am.
I
am
I'm
happy
with
the
way
that
he
first
class
got
like
the
end
of
that,
because
we've
got
more
confusion
than
then
then
solution.
For
this
thing
we
should
just
get
stick
with
the
name
right,
but
yeah.
I
I
I.
D
D
D
A
C
A
Okay,
also,
like
I
said
it,
I
think
if,
if
I
assign
this
to
myself,
I
think
it
will
help
me
with
because
I
still
I'm
still
fuzzy
on
it.
Like
you
said
you
implemented
it
and
it
was,
it's
still
sometimes
confusing.
So
I
think
it
might
just
be
helpful.
Okay,
I'll
we'll
go
ahead
and
assign
this
stuff
sign
this
one
to
me.
A
D
A
A
A
A
D
I
mean
it
would
be
good
if
they
can
do
some
if
they
can
check
if
the
inject,
if
in
the
gynex.com
file
generated,
is
this
one
yeah
yeah
it's
there
following
he's
saying,
doesn't
get
created
with
global?
Oh,
it
doesn't
get
created.
Okay,
okay!
I
would
have
to
check
that
and
see
if
I
can
really
use
that.
So
if
someone
wanna
take
a
look
into
that
and
try
to
reproduce
at
least
it
would
be
helpful.
D
A
A
D
Let's
do
this,
so
you
can
assign
both
to
me
and
you
maybe,
and
anyone
that
want
to
join
us
and
we
can
and
we
can
try
at
least
to
reproduce
that
we
can
say
hey.
We
just
need
someone
to
try
to
reproduce
that
and
and
and
or
we
can
just
because
I
I
don't
want
to
be
the
I
mean
the
git
keeper
is
the
wrong
word,
but
I
mean
the
the
the
show
point
for
all
of
those
issues
right
but
yeah
to
me.
D
It
works
better
if
I
can
reproduce
other
than
just
read
the
issue,
because
I
can't
by
my
head,
my
head
just
know:
what's
going
on
it's
all
good,
so
we
can
we
can
we
can
do.
Maybe
what
we
can
do
is
actually
doing
the
same
thing
that
ccli
does
and
on
set
up
once
a
month.
We
can
just
do
some
bug
scrub
like
let's
bootstrap
a
cluster
and
try
to
reproduce
those
together
and
see
hey.
This
is
something
that's
really
happening.
Probably
this
is
the
part
of
the
code.
What
do
you
think
about
that?
A
D
Got
a
better
vision
on
that
like
related
issues
and
he's
good
at
that
right,
so
we
can
talk
with
him
and
say:
hey.
You
know
we
are
thinking
about
doing
this
in
a
group.
So
what
do
you
think
about
electing
one
or
two
issues
and
saying
hey?
We
can
try
to
do
this
bug
scrub
over
that
issue
together
and
he
knows
yeah
yeah
we
can.
We
can
talk
with
him.
A
Okay
yeah
this
one
I
was
looking
at
earlier
today.
He
said
the
helm
chart
can't
be
installed
and
he's
doing
an
upgrade.
So
I
asked
him
what
version
he's
upgrading
from,
because
I
did
a
direct
upgrade.
I
did
a
direct
installation
with
for
413
and
it
worked
just
fine
and
long
was
reporting
the
same
thing.
A
D
A
Next
yeah,
it
didn't
do
that
when
I
used
the
I
use
the
bot.
Okay,
all
right.
This
is
an
interesting
one.
I
was
looking
at
this
one
earlier
today
and
I
wanted
to
get
your
thoughts,
ricardo
and
maybe
anybody
else.
It
looks
like
when
the
controller
gets
shut
down.
It
waits
for
10
seconds,
I'm
guessing
that's
waiting
for
the
pod
to
exit
gracefully.
A
B
C
D
Yeah
yeah
and
this
can
be
configured
configurable.
Yes,
I
think
so.
So
what
what's
happening
is
that
what's
exactly
happening
there?
Is
that?
Because
the
ingress
controller
process
is
the
main
process
right,
he
don't
want
that
to
exit
before
he
don't
want
that.
We
we
don't
want
that
to
exit
before
the
ingen
x
stop.
So
this
is
why
we
kind
of
weight
right.
D
A
C
C
A
E
D
Do
that
a
lot
of
time,
if
james,
if
you
cannot
assign-
because
I
don't
know
if
if
aditya
is
a
member
of
the
work,
so
what
did
you
you?
Can
you
can
jump
into
the
the
issue
and
do
an
assign
for
yourself
if
the
bot
fails,
because
the
bot
only
assigns
people?
That's
name,
that's
a
member
of
the
organization.
Okay,
yeah!
That's.
C
C
C
B
A
A
A
D
D
C
A
A
C
A
A
A
D
Let
me
ask
you
to
take
a
look
into
that
on
the
next
meeting.
Okay,
this
is
yeah
because
we
are
now
we
are
just
focusing
on
the
open,
telemetry
stuff
right,
so
I
wouldn't
say
that
we
should
right
now
spend
too
much
time
on
open
tracing,
at
least
as
soon
as
we
can
get
the
proper
scenario
between
the
electrician
of
the
telemetry
and
things
that
we
need
to
change.
D
D
He
he
just
did
the
split
between
the
containers.
So
now
we
have,
we
have.
There
is
an
ongoing
pr
that
I
was
reviewing
and
now
we
may
build
like
the
open,
telemetry
container
on
a
separate
container
of
the
nginx
container.
So
this
is
working
fine
and
I
just
want
to
take
the
same
approach
for
the
security
thing.
It's
gonna.
D
It's
gonna
reduce
a
lot
of
our
build
times
and
and
after
that
and
and
he
was
just
expecting
that
to
get
married,
I
just
need
him
to
fix
some
some
things
and
then,
after
that
we
he's
gonna,
start
working
on
the
open,
telemetry
siding
in
gynex
and
other
stuff.
Knowing
that
we
can
use
open
telemetry
in
a
different
configure.
A
A
I
think
that
would
just
be
helpful
from
our
reading
perspective.
Anybody
else
think
some,
that's
a
dumb
thing.
Just
let
me
know
moving
moving,
what
moving
the
actual
issue
to
the
top
of
the
issue
template
and
then
all
of
the
providing
information,
otherwise
you
have
to
scroll
through
and
just
have
to
figure
out
what
the
actual
issue
is
a
validating
web
hook
when
trying
to
deploy
a
gce
ingress,
the
ingress
class
annotation
seems
to
be
ignored
and
the
following
issue
happens.
A
D
D
C
D
A
A
D
D
Yeah,
I
I
so
yeah.
This
is
something
that
I
I
really
yeah.
I
really
wanted,
like.
I
know
that's
beginning
of
the
year,
so
probably
a
lot
of
folks.
They
are
on
on
vacations
and
ptos
and
etc,
but
we
need
we
need
some
other
folks
to
take
a
look
into
the
issues
as
well,
because
I
I'm
not
having
enough
time
to
look
at
both
of
them
and.
C
A
D
We
can
we
can
go
through
them
just
to
to
to
make
sure
that,
like
I'm,
just
keeping
the
dogs,
the
dogs
and
the
deployment
ones
and
leaving
to
jintao,
long
and
and
noah,
I
am
not
just
if
there
is
a
label
like
area
docs,
I'm
just
keeping
sorry
folks
go
to
the
go
to
that
one.
Eight:
zero!
Nine!
Eight,
please
james,
that
that's
the.
E
D
Yeah,
so
this
is
a
book
that
that
someone
found-
and
I
that's
that
one
of
the
inginx
reloading
this
just
needs
to
get
merged,
but
go
to
the
bottom.
You
can
yeah.
The
testing
is
failing
because
the
the,
how
can
I
say
the
code
base
is
the
hack
yeah.
I
have
asked
them
to
run
hack
blood
and
they
just
commented
on
the
request,
but
yeah.
We
need
to
to
run
that.
So
I'm
just
thinking
about
maybe
just
merging
that
and
then
fixing
that
code
base
in
another
pr.
D
Is
is
right
and
they
are
not
changing
anything
on
that.
What
happened
is
that,
because
this
is
the
first
pr
of
the
year,
probably
something
god
screwed
on
on
the
generation
on
the
auto
generation
themes
right.
So
that's
not
that's
not
their
fault.
So
I'm
just
thinking
about
fixing
that
and
and
maybe
forced
merging
by
hand
that,
like
approving
margin
and
then
fixing
that
that
thing,
but
I
am
not
sure
why
this
is
happening
actually
yeah.
That's
that's
from
the
generate.
A
D
So
this
is
one.
This
is
one
that
we
need
to
approve.
I
will
take
a
look
probably
later
and
check
why
and
fix
that
that
thing,
but
that's
mostly
about
reloading
it's
we
are.
We
were
missing
some
comparisons
to
see
if
the
new
configuration
was
equals
to
the
old
one,
so
config
wasn't
getting
reloaded
when
they
changed
the
like
some
certificate,
so.
E
D
D
D
I
added
some
comment
on
that
that
I
am
partially
okay
with
that,
but
I
am
a
bit
afraid
of
adding
that
could
allow
someone
to
abuse
that
direct
the
directive
to
something
that
we
don't
desire
like
running
a
lua
script
before
something
else
and
using
that
as
a
cve,
whatever
right
so
jintao
was
already
on
that
as
well.
I
just
I
was
as
my
point.
I
was
just
just
worried
about
someone
using
that
for
the
evil
and
I
need
I
and
I
need
people
to
think
with
that.
D
Just
to
say,
hey,
your
concern
is
valid
or
no
hey.
It's
just
okay,
don't
be
a
jerk
and
approve
that
okay,
so
I
I
just
need
someone's
point
of
view
as
well
on
how
if
someone
can
abuse
that
for
for
something
evil.
What
this
pr
does
actually
is
just
allowing
someone
to
to
add
into
the
base
image
a
script
that
gets
loaded
before
our
scripts
yeah.
D
Yeah
I
mean
I,
I
understand
the
point
I
am.
I
am
not
opposed
to
approving
that
I
just
need
someone
to
say:
hey
don't
be
worried
because
this
cannot
be
used
or
it's
really
hard
to
use
that
as
a
cve
point,
but
I
can't
say
that
with
confidence
that
this
is
true
or
not,
because
I
don't
understand
lua
too
much
to
say,
hey,
okay,
this
is
a
problem
or
not
right.
So
I
need
someone's
some
some
someone
point
of
view
on
that
as
well.
C
A
D
D
I
need
to
take
a
look
into
my
comment
on
october.
24
was
something
like
hey.
I
need
some
lua
reviewer
to
help
me
on
this,
because
I'm
not
sure
all
of
the
implications
of
that
right.
So
when
I
know
there
is
something
simple
in
lua
that
I
can
say,
hey,
okay,
this
is
fine.
I
usually
prove
those
that
may
change
the
way
that
we
do
some
sticky
sessions
or
something
like
that.
That
may
break
things.
I
prefer
that
some
I
do
take
a
look
into
that
with
me.