►
Description
[SIG-Network] Ingress NGINX Bi-Weekly Meeting for 20220707
A
Hello,
everyone,
my
name,
is
james
strong.
This
is
the
july
7th
meeting
for
the
signet
looking
sub
project
ingress
in
genex,
which
is
a
under
the
cncf
code
of
conduct,
which
means
be
kind
to
each
other.
If
you
have
any
issues,
please
go
ahead
and
report
those
to
us
order.
The
cncf
with
that
we've
got
one
new
person
on
the
call.
A
I
thought
we'd
have
a
couple
more
from
the
conversations
we
had.
If
you
are
feeling
up
to
it,
go
ahead
and
introduce
yourselves.
A
C
Hi
james,
I
am
terry-
I
am
just
new
to
the
open
source
so
recently
I
started
this
so
yeah.
I
am
interested
in
networking.
So
that's
why
I'm
interested
in
joining
this
school.
A
I
just
posted
to
the
chat
the
the
doc
that
you're,
seeing
that
I'm
sharing
right
now,
all
right
so
ricardo
we've
got
about.
You
said
you've
got
about
10
minutes.
So
do
we
want
to
skip
to
like
the
stabilization
project
discussions
and
then
we
can
go
through
issue
triage.
No
just
one.
Second,.
B
So
steve
asked
if
he
needs
to
raise
hands
to
speak,
hey
steve,
welcome
you
don't
need
to
you
don't
need
to
unless
someone
is
already
speaking.
So
if
you
want
to
speak,
go
for
it.
E
Right
five
minutes,
so
this
is
so.
I
was
pretty
particularly
interested
in
this
networking
area.
So
this
is
my
first
time
you
know
contributing
to
kubernetes.
I
think
so.
It
has
been
easier
to
contribute
now
right.
So
back
in
2008
it
was
a
little
tougher
as
far
as
I
know
so,
yeah
and
just
one
question,
so
I
have
seen
this.
You
know
kubernetes
release
leave.
E
A
It's
part
of
the
release
team
for
sig
release.
That's
the
person
who
helps
run
a
major
release
so
making
sure
that
things
are
getting
completed
and
ran
on
time
and
working
with
all
of
the
cigs
to
make
sure
things
are
getting
completed.
So
it's
part
of
a
shadow
program
for
the
sig
release
team,
so
you
can
apply
to
be
a
sig
release,
shadow
and
there's
other
jobs
within
the
sig
release
team
that
you
can
also
apply
to
shadow.
A
For
you
know
this
is
a,
I
would
say
you
know
helping
out
with
this
project,
understanding
how
we
do
releases
and
then
understanding
how
the
kubernetes
team
does
releases
as
well
is
helpful.
But
this
isn't
the
sig
release
meeting.
They
have
community
meetings
like
this
as
well
that
you
can
join
and
you
can
understand
what
they're,
how
they,
how
they
work
and
how
they
accomplish
things.
But
this
is
this
is
a
separate
sub
project
for
the
ingress
engine
x.
Meeting
from
the
networking
perspective.
A
Awesome,
I
know
we
had
a
couple
other
people
and
steve.
I
can
I'll
drop
it
in
the
on
our
dev
channel,
a
little
bit
more
resources
to
help
get
started,
but
for
right
now
we're
working
on
we're
working
through
a
project
to
help
with
stabilization
of
the
ingress
engine
x
project.
So
that
includes
a
couple
things
right.
A
So
there's
some
things
that
we've
noticed
that
we
need
to
fix
that
we've
added
to
the
project
and
we're
not
really
accepting
new
feature
requests,
except
for
ones
that
we
deem
are
necessary
for
the
project.
So
that's
what
we're
going
to
do
an
overview
right
now,
unless
leon
wants
to
introduce
himself
as
well
we're
going
to
go
ahead
and
get
started
on
that.
A
Okay,
so
since
this
is
new,
should
do
we
just
want
to
run
through
the
tickets
that
are
on
here
making
sure
that
folks
aren't
blocked
on
things?
What
do
you
guys
all
think
I
I
I've
just
been
working
through
and
closing
them
as
I've
been
completing
things.
B
No,
I
so
the
last
thing
that
I'm
doing
right
now
is
sorry
working
on
on
the
leader
election
stuff.
I
think
this
is
the
only
one
blocking
us
for
the
next
release,
so
I
did
some
review
on
on
gentile
pr
on
on
that.
I
I
think
it's
fine.
I
just
wanted
to
make
sure
that
we
have
some.
B
F
James,
there
are
two
issues
I
think
I
added
the
stabilization
tag
on
them,
but
they're
not
in
the
project.
One
is
that
we
didn't
know
until
now,
but
it
seemed
metrics
are
broken
and
somebody
very
kindly
and
helpfully
agreed
to
work
on
it.
So
if
you
look
at
the
tags,
stabilization
one
is
for
matrix
other
one.
Is
this
vulnerability
for
curl
and.
A
Is
this
the
vulnerability?
Is
this
the
one
I
posted
about
in
the
dev
channel
the
other
day.
E
G
D
B
So
I
I
need
to
drop
for
like
five
minutes
and
then
I
will
be
back,
but
I
think
that
for
the
vulnerability
stuff
as
soon
as
we
do
the
new
release
with
gentile
changes,
we
do
an
update
on
on
the
alpine
linux.
So
it
works
fine
for
the
metrics.
I
think
it
would
be
good
if
we
can,
if
we
can
fast
try
to
see
if
there
is
a
solution
on
that
on
version
1.
B
F
Okay,
okay,
so
james.
If
you
see
the
last
comment
on
this,
this
matrix
one,
there
was
a
earlier
pr.
Somebody
had
worked
on
matrix
and
the
very
last
comment
he
said:
he'll
take
a
look
at
it.
We
don't
know
if
it
will
be
in
time
for
our
intended
release,
but
I
think
we
should
just
add
this
to
the
project
distribution
project
and
then
wait
and
see
how
it
goes.
A
F
A
Okay,
well
yeah!
No,
that's
I
mean
what
bug
fixes
need
to
be
in
so
I
have
no
issue
with
that
as
far
as
if
it
gets
into
the
one
two
two,
that's
a
different
story:
right,
okay,
so
I
added
that
long.
I
probably
should
add
you
to
to
the
project
stuff.
Can
you
follow
up
with
with
gentile
and
figure
out
what
pull
requests
you
need
to
put
in,
so
you
can
have
access
to
the
project
as
well?
A
Like
that
any
of
the
those,
but
I
will
see
test
the
project
labeling
and,
if
not
then
go
ahead
and
put
it,
follow
up
with
the
pull
request
and
get
that
fixed
okay,
so
I
think
this
one's
ready,
the
update,
ingress
engine
x
and
base
in
one
place.
A
I
think
I
just
need
somebody
to
re
finish
reviewing
that
gentile.
I
know
you
were
looking
at
it.
It
builds
it
works.
A
F
A
A
No,
it
looks
like
gentile's
having
issues
so
we'll
just
move
on
to
the
next
stuff.
B
A
I
wanted
to
ask
about
the
open
telemetry
stuff
and
where
we're
at
I've
been
I'm
not
following
it
100,
but
is
there's
there's
two
open:
pr's
isn't
there?
Where
are
we
at
with
that
long?
Do
we
are
we
able
to
get
that
one
finished
and
ready
for
the
the
one
two
two
release.
F
So
the
summary
of
that
is,
we
need
to
see
when
ricardo
will
be
free
and.
F
Image,
that's
done
closed,
no
more
in
the
queue
anywhere.
That's
done
so
now.
As
per
what
ricardo
said,
the
plan
for
doing
it
in
stages,
one
is
to
just
get.
It
enabled
get
the
open,
telemetry
module,
enabled
compiled
and
enabled
so
that
when
we
launch
the
controller
it's
available
and
the
libraries
and
the
module
is
available
that
part,
so
the
work
work
actually
is
just
we
wait
until
ricardo
is
free
to
have
a
discussion
on
that,
but
I'll
manage
that
there
is
no
like
stopper
on
that.
F
Only
thing
is
when
the
card
is
free,
daniel
and
me
have
to
kind
of
kind
of
explain
how
we
are
loading
right
now.
It's
a
manual
loading
so
from
there
ricardo
said,
he'll
do
a
testing
so.
D
B
Okay,
I
can.
I
can
try
to
to
make
some
time
during
this
week
or
next,
even
if
we
don't
have
the
ingredients
next
meeting.
Just
so
you
folks
can
can
work
on
that
with
me.
Okay,
so
that's
that's!
That's
fine!
Let's,
let's
see
if
I
can
find
something
like
30
minutes
next
week
and
as
soon
as
it's
fine
for
you
and
and
the
folks
working
on
that
I
can
I
can.
I
can
make
some
time
and
take
a
look
into
that.
Let's,
let's
just
just
don't
don't.
Let
me
forget
on
that.
F
Okay,
the
reason
I
asked
about
your
time
today
is
that
it
one
there
was
a
performance
issue
going
on
and
there
was
like
not
enough
information
or
analysis
and
all
that,
but
suddenly
now
there's
an
update
on
that
and
he's
got
a
code
up
and
he's
saying
he
doesn't
know
how
to
use
gdb
on
it.
So,
but
it's
a
kind
of
three
or
four
people
have
like
and
joined
on
that
issue.
Saying
yeah
performance
drop,
I
don't
know
if
you
want
to
address
it
now.
F
I
want
to
take
it
up
later,
but
I
think
it
should
be
marked
for
stabilization.
Just
for
you
know,
reviews
whenever
we
get
time
there's
a
this
is
a
separate
issue
from
open,
telemetry,
yeah
yeah.
This
is
this
is
basic,
pure
performance,
not
not
ready
to
open
to
limit
it's
a
performance
revolution.
A
F
I
mean
the
thing:
is
it
was
hard
yeah,
it's
eight
one,
six,
six.
A
A
Who
why
do
you
want
me
to
sign
this
one
out
to
you?
Are
you
testing.
F
Oh
yeah,
you
can
assign
to
me,
but
it's
actually
way
beyond
my
head,
as
in
okay,
he's
the
guy
who
actually
very
the
last
person
who
commented
on
this
he's
done.
Some
lot
he's
done
some
work
that
is
kind
of
helpful,
so
he's
got
a
co-dump
and
he
says
he
doesn't
know
how
to
use
gdb
against
it.
B
Yeah,
so
the
cardone
would
be
something
that
we
should
probably
interact
with
nginx
folks
or
within
tao,
because
I
know
the
gentile
loves,
sequel
and
yeah
okay,
so
we
may
we
may
do
that.
I
I
have
no
idea
what
I
can
do
with
gdp
other
than
usually
I
just
erase
the
cordons
from
my
machine
because
they
take
too
much
space.
F
F
So
tying
into
this
comes
back
to
k6,
so
we
did
half.
We
talked
about
it
when
we
know
about
it,
but
this
come
up
a
couple
of
aspects.
So
whenever
we
get
time
either
now
or
asynchronously,
we
need
to
close
a
loop
on
one
question
so
that
we
got
to
basically
we
need.
We
need
tls,
enabled
when
we
run
this
k6
right.
So
I
got
a
free,
nom
free
domain
and
I
put
in
a
let's
encrypt
thing
and
I
was
asking
james
if
that
is
okay,
so
I
don't
know
how
we
want
to
do.
A
B
So
even
there
we
could
take
a
look
into
the
times
of
the
end-to-end
tests
and
see
if
we
have
a
pattern.
F
On
that,
so
one
interesting
information-
this
guy
performance
guy
is
pointing
out
is
that
it's
not
the
controller
process,
but
the
worker
process
of
engine
x,
which
are
kind
of
so
it
could
be
red
herring.
Maybe
he's
got
he's
under
attack
or
you
know,
he's
got
usage,
but
the
whole
thing
is:
do
we
need
to
spend
time
now
or
do
we
like
first
clean
up
and
do
other
things
first
and
then
come
to
this.
A
Yeah,
I
mean
that's,
that's
the
thing
right.
We
have
to
prioritize
these
as
we're
working
through
these
with
the
issues
in
the
project.
So
if
we
look
at
inside
the
project
that
we're
looking
at,
let
me
just
I
right:
we've
got.
A
A
A
These
are
ones
that
are
in,
or
people
are
working
on
right
now,
the
release
one's
got
a
lot
more
stuff.
F
A
We
need
to
get
a
list
of
those
long.
Can
you
pull
that
together.
A
So
the
the
the
cves,
the
two
cve
issues
that
we
have
well,
those
will
get
taken
care
of
when
we
do
the
new
release.
So
are
we
gonna
we're
gonna,
wait
for
the
124
stuff
to
do
the
one
two
two
release
right,
gentile.
G
F
G
A
Okay,
okay
sounds
good,
we'll
do
that
then,
so
I
just
want
to
make
sure
that
the
work
that
we're
doing
we
want
to
keep
track
of
it
and
put
it
in
progress.
So
as
things
as
folks
are
working
on
things,
let's
make
sure
that
they're
tracked
here
and
in
progress
so
long.
If
you
can't,
if
there's
anything
on
there,
that
you
can't
you
can't
move
in
or
do
anything
with,
just
let
me
know
and
we'll
get
that
figured
figured
out
and
taken.
So
the
open
telemetry
is
on
the
board,
which
is
good.
A
A
F
F
C
A
F
A
Okay:
let's
get
that,
let's
get
the
let's
get
the
base
image
update
in
one
place
merged
in,
so
that
when
you
do
do
this
for
the
next
time,
it's
just.
We
have
that
easy.
So
I'll
get
that
in
I'll
open
up
an
issue
in
a
pr
I'll,
open
up
the
issue,
and
then
you
can
just
tie
it
to
the
pr
when
you're
done
when
you,
when
you
open
it
up
and
we
can
test
316.
A
Gentile
you
think
next
week
we
can
do
the
do
the
one.
The
one
two
two
release.
F
Cool
sounds
cool.
One
other
aspect
of
that
316
is
that
remember
you
posted
that
you
message
right
that
we
need
to
upgrade,
but
there
that
vulnerability
is
only
in
304,
okay,
open,
ssl
or
whatever
library.
That
is
it's
only
one
specific
version
of
that
library
304.
So
we
are
not
there
yet.
I
see
our
current
controller
and
basement
doesn't
use
it
so
yeah.
Yes,
that's.
F
A
If,
if
we
can
run
316,
let's
just
upgrade
to
alpine
316.
got
it
okay,
okay,
I
was
talking
with
ricardo.
Yesterday
too,
I
think
I
can't
remember
what's
the
latest,
is
it
once
we're,
not
one
on
120
right?
He
wants
to
look
at
121.,
121
yeah
we're
going
to
look
at
121
as
well
yeah.
So
that's
probably
not
gonna,
be
on
the
one.
Two
two
one,
two
two
release.
F
F
So
actually,
there's
a
couple
of
them,
but
eight
seven,
two
five
should
be
a
good
peak
and
eight
seven,
four
three,
so
they
both
so
eight
seven,
four,
three,
eight
seven,
two
five
and
eight
six,
eight
zero.
All
three
are
talking
about
the
same
thing,
which
one
give
me
the
numbers
again.
Sorry,
eight,
seven,
four
three
and
the
second
number
is
eight
seven.
Two
five
yeah
I've
got
two
five.
Okay,
I
got
eight
seven,
four,
three,
eight
seven,
two
five
and
the
third
one
is
eight
six.
Eight
zero.
F
F
A
Okay,
I've
got
that
in
there
and
then
we'll
close
and
consolidate
as
needed,
use
one
for
the
tracking
and
then
I'll
add
add
to
the
project.
A
Okay
sounds
good,
so
we've
got
good
timing
on
these.
These
three
things
for
next
week's
one,
two
two
release
and
we'll
re-prior.
We
can
re-prioritize
these
once
we
get
that
release
completed
so
for
our
next.
Our
next
meeting
we
can
discuss
who's
going
to
work
on
probably
the
car's
going
to
work
on
the
control.
Plane's
fix
these
as
needed
for
long
you'll
have
a
list
of
the
pr's
that
need
to
be
added
that
need
to
be
completed.
A
I
will
I
want
to
review
this
one:
okay,
okay,.
F
We
do
need
to
ask
them
for
the
flame
or
something
just
see.
The
chat
in
zoom
chat
said
that
for
that
one,
the
performance
one,
you
need
to
ask
them
for
something
called
flame
chart.
F
A
Sounds
good
we've
got
these.
We've
got
those
listed
out
before
I
move
on
to
the
triaging
some
issues.
A
E
E
A
A
E
F
E
A
Go
look
at
all
this.
Oh
wow.
A
A
A
A
I
think
the
controller
does
create
endpoints
and
since
we're
not
using
endpoint
slices,
this
would
probably
be
a
good
good
one,
for
I
think
this
would
be
a
good
first
issue
if
we
go
look
to
see
where
we're
creating
the
endpoints
and
move
that
move
it
to
endpoint
slices.
F
Let
me
I
don't
even
know
what
endpoint
slices
means,
but
he's
talking
about
implementation,
which
means
an
alternative
to
the
english
object
routing
to
so
backend
services.
F
E
A
A
Endpoint
slices
just
allow
you
to
push
only
the
the
pods
for
that
node
as
an
endpoint,
instead
of
getting
the
whole
list
of
endpoints
you'll
just
get
a
specific
subset
of
the
endpoints
that
are
running
on
that
node.
So
that's
the
the
way.
I
understand
the
difference.
So
if
you
have
a
hundred
endpoints,
you
won't
get
all
100
as
a
node.
You'll
only
get
the
ones
that
are
running
on
your
on
your.
A
F
It
says
the
feature
is
not
enabled
for
in
what
whatever
the
bot
for
for
this
project.
That's
what
zinto
was
saying,
and
he
put
it
he'll
put
in
a
pr
to
get
it
enabled
right,
jinta
yeah,
that
tag
called
project,
slash
project,
it's
it.
It's
still
not
working
for
this
project.
You
have
got
to
put
in
a
pr
to
make
it
work
right.
C
F
So
they
want.
They
want
the
project
to
include
the
json
and
but
the
whole
problem
with
grafana
and
even
any
other
collectors
is
that
there's
a
it's
always
maintained
out
of
out
of
net.
You
know
a
lot
of
these
software
trees.
It's
always
kept.
If
it
is
dashboard,
then
it's
hanging
off
of
the
grafana
website
and
this
and
the
grafana
software
itself,
along
with
prometheus,
comes
from
the
operator
or
the
maintenance
of
grafana
and
so
on.
So
and
unless
there
is
a
very
commercial
curated
and
targeted
product.
F
No,
that
that
could
be
that
could
be
worked
on.
I
think
we
could
change
things.
The
problem
is
we
can't
maintain
that.
So
I
mean
it's
not
so.
If
something
changes
upstream
and
labels
change
or
something
happens,
then
we
have
to
maintain
and
keep
in
sync.
So
it's
like
two
places,
the
same
software
being
maintained.
So
that
is
my
my
point
of
view,
but
I
don't
we
can
people
can
have
different
points
of
view.
A
Okay,
I
need
to
review
this
one
a
little
bit
more
because
more
than
the
time
that
we
have
on
here,
because
I
don't-
I
don't
understand
how
we're
deploying
how
we're
deploying
the
it
now.
I
understand
what
you're
saying
with
the
the
json
config.
F
Now
I
think
we
don't
deploy
it
by
default.
We
just
have
documentation
that
if
somebody
wants,
then
they
can
use
a
isolated
grafana
pair
in
that
same
namespace,
and
we
also
put
in
documentation
where
somebody
can
have
their
own
existing
grafana
and
prometheus
elsewhere
in
another
namespace,
but
then
they
can
deploy
a
service
monitor.
A
A
A
I've
got
hey,
there's
an
issue
opened
by
me:
that's
a
euro!
That's
funny!
I'm
gonna!
Look
at
that!
So
I'm
trying
to
work
through
the
oldest
ones
we
have
so
we
can
get
to
the
top
of
the
queue
with
fresh
ones.
So,
let's.
A
Well,
10
people
wanted
this.
If
someone
wants
to
pick
up
this
feature,
that
would
be
helpful.
That'd
be
great.
E
Yeah,
so
for
developing
kubernetes
right.
So
what
are
the
minimum
specs?
So
when
I
was
working
on
airflow,
they
have
given
the
minimum
specs.
So
each
12
gb
of
ram-
and
you
know
a
full
processor
cpu-
is
enough
for
developing
kubernetes.
A
You
know
what
I've
never
that
you're
you're
the
first
person
probably
asked
me
that
question.
If,
if
that's
enough
for
kubernetes
development,
that
should
be
enough
for
ingress
development,
because
a
lot
of
the
work
we
do
is
when
we
deploy
it
and
generate
a
new
controller,
we
deploy
it
to
kind
a
local
kind
cluster.
A
A
So
this
would
be
a
good
one
too.
So
I
think
the
the
code
is
there
for
the
change.
They
just
don't
have
any
end-to-end
tests.
So
I
can
also
slack
you
and
show
you
how
we
do
in
the
end-to-end
test
as
well,
and
if
you
can
get
the
end-to-end
test
added
that'd
be
a
that'd,
be
helpful.
So
you'll
probably
need
to
rebase
this,
because
there's
lots
of
changes
because
there's
an
opr
rebase
it
and
then
add
the
end
to
end
tests
and
then,
if
ci
passes,
we
can
get
that
pulled
in.
A
Okay,
awesome
all
right
folks.
Well,
we
got
we
got
through
three
issues
of
the
255.
Well,
not
255.
I
don't
know
how
many
are
not
triaged,
so
we'll
just
keep
going.
So
I
think
this
is
how
we'll
just
you
know
we'll
look
at
the
project
updates
where
we're
at
we'll
look
at
new
issues
and
see
if
they
need
to
be
added
to
the
project
and
then
go
through
and
triage
older
issues.
A
I
think
that
works
well
until
we
get
through
the
six
months
and
that
way
we're
not
looking
at
the
new
pr
and
then
we'll
get
through
the
prs.
Once
long
has
a
list
of
the
new
prs
of
the
prs
that
are
completed,
that
we
need
to
get
in,
but
we're
not
accepting
new
ones.
So
we
won't
be
reviewing
feature
requests
from
that
perspective
and
long
put
in
the
feature,
request,
portal
or
issue
template
the
fact.
The
the
the
discussion
around.
Why
we're
not
accepting
feature
requests,
so
I
think
we're
good
from
that
respect.
E
Jim
james,
so
there's
a
small
mistake
here,
so
I
have
mismatched
him.
So
could
you
just
change
it?
So
I
have
kept
the
new
name
in
chat,
so
it's
just
yeah.