►
From YouTube: [SIG-Network] Ingress NGINX meeting 20210803
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
Okay,
hello,
everyone
welcome,
so
this
is
sig
network
english
engine
exit
project
meeting
today
is
august
3rd
2021.
B
Yeah,
so
I'm
a
student
and
I
have
just
started
full
stack
development.
I
was
working
with
docker
files,
some
time
back
and
I
wanted
to
explore
kubernetes,
and
so
I
thought
you
know
to
join
the
community,
learn
and
work
together.
B
C
Yeah
hi
I'm
vomit
and
I'm
also
joining
this
meeting
for
the
first.
A
C
Yeah
hi
this,
like
I'm,
also
joining
for
the
first
time
brother
and
I
just
started
a
few
weeks
earlier,
with
the
ingress
and
yeah
working
with
long
and
everyone
else
yeah.
That's
it.
D
Hi
this
is
devraj,
so
I
am
also
new
to
this,
so
I
have
been
also
working
with
long
and
everyone
else
looking
forward
to
it.
Thank
you.
A
Cool
yeah.
We
have
a
lot
of
new
folks
today,
cool
amazing.
If
someone
else
wants
to
say
something,
please
feel
free
to
open
the
microphone.
This
is
really
a
really
soft
medium.
So
yeah
don't
be
shy,
okay,
cool!
So
we
we
have
some
recurring
topics.
I'm
gonna!
A
So
from
our
last
meeting
I
did
some
follow-up
on
the
on
the
ingress
giant
expression,
one
which,
if,
if
someone
also
can
take
notes,
I
am
really
bad
writing
and,
and
talking
so
just
feel
free.
Otherwise
I
can.
I
can
feel
the
things
that
I'm
saying
later,
but
so
I
promised
to
james
and
the
other
folks
that
that
would
finish
the
v1
theme
and
release
a
v1
better
one
which
we
did
last
week
so
now
ingress
v1
is
on
version
is
on
better
one.
A
Today
we
we
tried
to
release
a
helm
chart
for
this
one,
but
for
some
reason
it
didn't
get
released.
So
I'm
gonna,
I'm
gonna,
talk
with
long
later,
because
I
think
that
I
know
what
was
the
problem,
which
was
something
related
to
me,
and
we
expect
to
maybe
by
the
end
of
this
week
or
next
week.
Release
are
the
better
two
or
the
ga
and
have
the
community
testing
the
v1
and
start
the
process
to
drop
the
support
for
the
v0.x,
which
support
kubernetes
versions
older
than
119..
A
We
have
some
action
items
into
this.
One
of
those
is
documenting
the
migration
process,
which
I
know
that
long
is
taken
a
look.
I
will
take
a
look
as
well.
My
biggest
concern
with
this
is
making
everybody
aware
that
that
v1
beta1
is
not
supported
anymore,
so
people
need
to
migrate
their
objects
on
on
kubernetes
storage.
A
So,
like
writing
the
object
into
v1
and
also
about
the
ingress
class,
which
changes
a
little
bit
for
the
majority
of
of
users,
is
just
adding
a
boolean
in
ingress
controller
to
watch
objects
without
ingress
class
and
also
create
an
ingress
class
at
the
full
ingress
class,
but
for
folks
having
that
have
more
than
one
ingress
controller
running
on
their
cluster
or
their
environment.
This
is
gonna,
probably
be
a
problem,
so
we
need
to
proper
document
that
any
any
any
questions
folks
about
the
viewer.
A
Oh
right,
okay,
cool.
The
next
item
in
the
agenda
is
triage.
So
noah
you
made
an
issue
triage
or
those
are
like
issues
that
need
our
attention.
E
A
mixture
so
the
the
first
one
in
the
list
was
also
in
the
in
the
list
of
the
last
meeting.
Maybe
you
can.
F
A
Yeah
yeah
all
right,
so
I
thought
this
one
was.
Let
me
take
a
look
because
I
thought
that
we
had
another
one
about
this
sorry,
so
kubernetes
version
113.
A
It
seems
that
this
this
panic
is
from
api
machinery
and
not
from
the
controller.
E
This,
I
think,
there's
a
comment.
E
E
D
C
A
A
Okay,
I
know
what
I
did
wrong
anyway.
Okay,
so
apparently
the
code
is
trying
to
extract
something
from
the
token
and
and
the
token
doesn't
have
this
this
value,
so
it
fails
and
they
have
posted
actually
a
patch
right,
so
they
could.
A
They
could
post
something
like
because
they
could
post
the
fix,
but
anyway
yeah
okay,
so
this
is
accepted.
We
can.
We
can
take
a
look
and
let's
mark
it
as
priority,
so
I
can
take
a
look
later
or
if
someone
want
to
take
a
look
into
this,
it
seems
pretty,
I
wouldn't
say
easy,
but
it
seems
pretty
pretty
clear
how
to
solve
it,
which
is
maybe
verifying
if
the
certificate
certificate
is
is
new
right,
because
apparently
it
works
here,
because
the
because
the
desert.
A
Isn't
new
but
the
certificate
is
new,
so
it
can
extract
the
the
certificates
here,
but
it
it
cannot
continue
and
it
fails
here.
So
probably
if
someone
want
to
take
a
look
into
this
is
just
putting
like
an
r
here
and
verifying
if
that
certificate
is
also
new,
okay,
cool.
Thank
you
noah
for
this
one
yeah.
So
let's
go
to
the
other
one.
E
A
F
I
don't,
I
don't
know
that
I've
seen
it
with
auth
url,
but
there
are
a
number
of
bugs
due
to
like
the
scope
of
where
an
annotation
applies.
If
an
annotation
applies
to
something
in
the
server
scope,
it
ends
up
getting
basically
randomly
picked
which
which
ingress
is
going
to
apply
the
setting,
and
you
can
have
all
sorts
of
different
bugs
crop
up.
Based
on
that.
F
F
A
G
I
saw
one
issue
presented
today
and
and
he's
basically,
and
that
issue
was
describing
a
use
case.
That's
basically
not
supported,
doesn't
make
it's
basically
a
misconfiguration,
that's
what
I
I
gathered
from
first
look
at
it.
I
don't
know
I'm
on
phone,
so
I
don't
know
if
this
is
the
same
issue
that
I
that
I
saw
today
or
yesterday-
god
just
it's
very
recent.
G
So
but
generally
I
was
saying
I
think
somehow
I
think
he
has
some
experience
and
I
was
thinking
of
working
with
him
on
the
at
least
more
than
one
auth
related
issues.
Okay,.
A
So
I
won't
forget
so,
and
you
won't
forget
okay,
but
feel
free
to
unassign
you
if
you
think
it's
not
something
that
you
like
you
you
wanna
deal
later
and
I
can.
I
can
take
a
look:
okay,
okay,
okay,
okay,
cool
next
one,
admission
controller
that
job
has
create
container
config
error,
azure
fireball
cd,
when
changing
the
configuration
of
the
using
giant,
shutter,
c
commission
job.
G
Quick
question:
broomage:
do
you
have
bandwidth?
Do
you
have
time
to
work
with
me
on
this
issues?
G
G
A
A
G
A
A
A
Okay,
we
need
to
check
this
because
there
is
a
run
as
user
2000
here
unless
that
user
is
removing,
is
removing
this
line,
it
should
be
working,
so
anybody
wants
to
to
take
a
look
into
this.
C
E
Is
if
if,
if
this
is
really
a
bug,
or
is
it
intended
to
work
like
this,
so
the
the
person
who
created
the
issue
just
asked?
If
it's
really
a
bug,
and
I
think
we
should
yeah,
we
should
at
least
reply
to
him
because
he's
he's
willing
to
to
give
it
a
try.
So
I
I
cannot
really
say
if
it's
a
bug
or
not.
E
A
A
E
Yeah
that
that
that's
mine,
so
you
you
gave
me
some
feedback,
okay,
I
I
will
go
through
it,
but
one
question
is
still
open
and
that's
how
we
should
proceed
with
ci
testing.
You,
you
told
me
something
well,
the
question
is
how
we
should
proceed
if
we
want
to
include
it
in
the
ci
pipeline,
so
you
said
we
have
to
manually,
create
an
image
and
push
it
somewhere,
but
so
I
I
have
no
credentials
to
push
it
on
the
on
the
gcr.
E
A
A
Worked
yeah
so
as
as
soon
as
this
gets
merging.
E
A
The
image
you
got
gets
into
it.
It
starts
a
process
which
is
club
view.
Let
me
let
me
see
if
I
can
probably.
A
Here
and
only
a
small
subset
of
folks
have
access
to
this,
because
that's
some
rules
of
kubernetes.
A
F
A
We
can
open
anyway,
I
usually
I
open
this
and
and
leaves
like
a
zoom
running,
so
people
can
think
of
it.
So
this
runs
and-
and
it
tries
to
compile
everything
of
these
images
and
generates
an
image
on
this
staging
right.
After
that,
we
we
need
to
promote
that
image
to
production,
and
this
this
promotion
is
another
pr
on
another
ripple.
It's
it's
documented
right
here,
but
is
it
here
on
the
release
and
the
this
first
step
right?
So
this
is
so
as
soon
as
it
gets
merged.
A
E
B
E
Okay,
fair
enough,
I
I
I
will
go
ahead
and
and
fix
the
proposed
proposals
you
made
and
then
we
will
move
that
forward.
Okay,
cool.
A
Okay,
can
you
clarify
exactly
those
tests
are
doing
see
like
they
just
only
did
in
general.
Five.
Is
that
correct,
yeah?
So,
okay,
so
at
least
one
thing
that
I
expect
is
when
we
add
something
new
into
the
engine
x
team
play.
A
We.
I
expect
that
we
test
if
this
doesn't
break
the
the
the
the
right.
So
if
something
some
scenario
that
we
add
into
the
the
the
gynex
configuration
doesn't
break,
and
it's
not
going
to
put
like
the
image
on
crash
loop
back,
because
if
in
this
case
we
are,
we
are
allowing
users
as
like
a
platform
owner.
We
are
allowing
users
to
create
their
own
ingress
object
and
if
they
are
and
if
they
put
something
that
that
breaks
all
of
the
ingresses
from
a
cluster.
A
You
are
gonna
impact
like
all
of
the
clusters,
and
then
you
are
gonna,
get
like
a
cv,
or
something
like
that
that
you
are,
you
are
able
to
do
a
denial
of
service
right
because
you
are
you
are
you
are
breaking
the
ingress
control
over
the
ingress,
the
nginx
from
from
the
whole
cluster?
So
what
I
expect
here
is
not
just
validating
like,
but
also
doing
a
simple
request
to
to
check
if
the
back
end
is
returning.
A
What
I
need
right,
so
I
am
adding
proxy
ssl
verify
equals
to
off,
and
I
I
need
to
to
call
that
ingress
and
receive
an
http
okay
and
then
I
I
add,
with
proxy
ssl-
verify
equals
to
one
and
some
secret,
and
I
still
need
that
to
to
come
back.
So
they
are,
as
as
this
user
is
adding
something
between
engine
x
and
the
back
ends.
We
need
first
to
verify
that
this
this
is
not
gonna
break
the
engine
x,
configuration
and
second,
that
this
is
gonna
work
properly
as
expected
right.
A
So
this
is
actually
what
I
expect
it's,
not
it's,
not
it's!
I
would
say
that
it's
simple,
it's
as
simple
as
like:
let's
go
to
a
test,
for
example,
of
an
annotation
or
something
like
that.
So
like
this
back-end
protocol,
so
the
user
added
a
new
annotation
that
you
can
have
like
an
automatic,
an
automatic.
F
A
Right,
I
guess
what
someone
in
this
meeting,
so
what
I
need
to
to
do
is
okay.
I
want
to
create
this
ingress.
I
want
to
make
sure
that
this
ingress
exists
and
I
want
to
wait
for
this
ingress
to
to
enjoy
next,
but
I
also,
I
also
wanna
to
make
a
test
and
say:
hey
okay,
so
when
I
call
thisbackendprotocol.food.com
it
returns
as
expected.
So
let
me
see
if
I
can
find
one
that
works
affinity
mode,
maybe.
A
Yeah,
we
need
to
fix
all
of
those
tests
and
make
calls,
but
where
is
it
here?
So
we
need
to
make
a
final
call
and
say:
yeah.
Okay,
this
works
as
expected,
and
these
returns
as
expected.
Even
if
we
are
not
very
fine
if
the
back-end
have
any
ssl
or
not
right,
but
we
need
at
least
to
say:
okay,
this
is
working.
This
is
not
breaking
users
got
it.
E
Okay,
maybe
it's
helpful
if
we
just
hint
hint
the
person
to
this
line
of
code
or
something
and
say
some
some
some
sentences
about
it.
F
E
Yeah,
let's,
let's
just
do
it
right
now,
maybe
because
you
already
have
the
the
yeah
this
one.
This
reference.
A
E
A
If
anybody
anybody
that
that
sees
some
e2e
test
missing,
this
call
also.
I
I
strongly
advise
to
to
add
those,
as
this
helps
us
not
only
knowing
that
the
configuration
is
there,
but
that's
also
working.
So
I've
opened
a
good
first
issue
this
last
week
with
some
examples
of
files
that
I
missed
this.
This
call,
but
just
feel
free
to
open,
or
maybe,
if
you
want
to
give
four
new
contributors,
this
opportunity.
A
A
A
A
F
Do
it's
a
little
more
complicated
because
grpc
errors
do
not
use
the
http
status
code,
they
have
a
grpc
specific
header
that
describes
the
error.
So
if
you
want
your
client,
your
grpc
client,
to
get
a
descriptive
error,
the
response
needs
to
include
the
the
headers
so
that
it
looks
like
a
grpc
error.
F
You
would
add
that
and
then
there's
the
way
that
the
grpc
encodes
error
messages
is
by
putting
them
into
an
httpg
header.
So
you
would
need
to
return
the
appropriate
header
to
describe
the
grpc
error.
A
Actually,
this
is
not.
This
is
not
us.
This
is
an
example.
It's
a
good.
It's
a
good
example
yeah,
because
for
from
my
perspective,
what
what
we
can
actually
do
is
like
create
a
comma,
a
common
file
for
all
of
those
errors
and
adding
them
when,
when
we,
when
we
detected
the
backend,
is
a
grpc.
I
didn't
then,
and
like
adding
this
file
into
that
location.
F
That
would
be
good
by
the
way.
A
go.
Grpc's
client
does
the
same
sort
of
translation
in
more
recent
versions
it
when
it
receives
like
a
http
401.
It
converts
it
to
a
grpc
unauthenticated
error
in
the
client
side.
So
the
same
sort
of
thing
I
don't
know
they
must
be
asking
for
it,
because
not
all
clients
do
that.
F
A
But
maybe
maybe
that's
something
related
to
their
application:
yeah,
okay,
so
they
they
also.
They
are
putting
some
examples
here,
like
server
snippets
for
those.
So
I
guess
like
we,
we
we've
got
like
the
the
problem
and
also
the
solution.
So
in
my
opinion,
it
would
be
something
like
we
can
add
this
as
a
common
file.
A
F
A
D
A
D
E
A
D
A
A
F
A
E
I
think
so
go
to
the
last
comment.
Yeah
so
long
or
already
tried
also
also
james,
but
I
think
james,
just
the
the
edits
to.
I
think
there
yeah.
A
A
Amazing
yeah
yeah
and
thank
you
thank
you,
everybody
think
of
everyone
that
that
attended
and
also
took
like
ownership
of
those
issues.
It's
a
really
really
helpful.
So,
probably
next
week
we
are
going
to
release
the
view
on
the
v100
beta,
beta
2,
or
maybe
the
ga
everybody
knows
where
to
find
us.
So
please
everybody
feel
free
to
reach
me
or
just
bring
problems
or
solutions
to
ingredients,
death
in
slack
and.
G
Small
chat
we
need
to
have
about
the
documentation,
because
I
think
the
scope
is
vast
and,
for
example,
the
first
two
lines
I
put
by
starting
to
attempt
making
attempts
is
first
line.
Is
that
blog,
so,
instead
of
copy
pasting
or
explaining
that
I
just
put
a
link
to
that
blog?
Is
that
how
we
want
to
do?
Or
do
we
want
to
like
get
descriptive
and
long
story
on
everything.
A
Yeah,
so
I
think
that
the
the
blog
post
is
is
good,
but
I
I
I
don't
think
that
the
blog
post
explains
for
the
user,
how
how
the
user
can
migrate
from
one
version
to
the
other
one
in
the
sense
of
ingress
class.
So
I
am
my
biggest
concern.
Right
now
is
about
the
english
class.
Sorry
change!
A
A
What
I
expect
long
right
right
now
and
we
can-
we
can
do
a
session
later,
but
I
will
just
have
like
the
weekend
to
do
this
right
now,
because
I
I
have
spent
all
of
my
upstream
contribution
time
yesterday
and
today,
but
what
I
expect
is
actually
we
we
can
write
some
scenarios
right.
So
user
is
on
version
blah
and
going
to
version
glove.
A
G
One
and
nine
yes,
what
I'm
thinking
is.
I
think
you
pretty
much
nailed
this
just
the
summary,
so
I
think
that
expands
to
having
three
or
four
different
use
cases
and
different
scenarios,
somebody
who
has
only
one
ingress,
somebody
who
has
multiple
and
needs
an
ingress
class
resource
you
know
created
and
all
that
that
seems
okay,
but
how
do
you?
How
do
we
build
up
to
the
whole
thing?
Is
medicine?
G
A
F
A
Of
experience
like
yeah,
we
can,
we
can
create
kubernetes
with
kind
version
116
cluster,
and
then
we
need
to
migrate
for
like
119
or
121,
and
we
need
to
migrate
from
from
ingress
in
chinax
zero
x
to
one.
I
guess
that
as
soon
as
as
we
start
to
make
those
experiments,
we
are
gonna
face
some
problems
that
probably
the
user
is
gonna
face
as
well
right.
A
So
we
need
to
pro
we
need
to
imagine
like
okay,
the
majority
of
the
clusters
that
we
have
around
is
like
folks
that
receive
their
own
cluster,
ready
from
google
or
from
amazon
or
like
they
are
using
kind
or
qbm,
and
they
don't
have
ingress
class
created
right
so
and-
and
I
don't
know
if,
like
on
the
helm,
chart
migration
that
ingress
class
is
going
to
be
created
or
not.
So
we
need
to.
We
need
to
exercise
that
got
it.
Okay,.
G
Okay,
so
basically
a
format
of
long
list
of
different
scenarios
right.
A
Yeah
yeah,
I
don't
think
I
don't
think
it
needs
to
be
like
a
huge
list,
but
I
think
that
we
should
we
should
at
least
try
to
cover,
like
the
majority
of
scenarios,
doing
exercising
ourselves
with
that
right.
I-
and
I
am
really
willing
to
help
that,
because
I
think
that
I
need
as
well
to
document
all
of
those
stuff,
so
you
can
like
create
a
hack,
md
or
something
like
that,
and
you
know
hackmd.