►
Description
[SIG-Network] Ingress NGINX Bi-Weekly Meeting for 20220510
A
Okay,
so
hello,
everyone-
this
is
signet
working,
resinji
nx
super
project
meeting
today
is
tuesday
may
10th
2022..
My
name
is
ricardo.
A
I'm
gonna
be
your
host
host
today,
and
I
would
like
to
kindly
remind
you
that
this
is
a
cncf
kubernetes
meeting
under
our
code
of
conduct,
which
is
basically
be
excellent
with
it
with
each
other
be
respectful,
and
if
someone
wants
to
raise
some
concern
about
the
code
of
conduct
here,
please
reach
a
code
of
conduct
on
slack
or
if
it's
not
something
with
me,
just
reach
me
on
on
slack
or
later
privately,
and
I
can
direct
you
better
or
conduct
that
kubernetes
dot
io.
A
A
We
don't
have
anything
right
now
we
do
have
some
action
items
from
before.
Maybe
not
so,
let's
take
a
look
into
some
issue
triage
and
also
I
will
open
my
my
spreadsheet
with
my
qe.
So
if
there
is
something
that
I
really
need
to
take
a
look,
I
will
add
here
and
move
forward
all
right.
So
let's
take
a
look
into
the
issues.
First,
okay,
anyone
wrong!
Do
you
have
any
any
of
those
that
needs
our
attention.
Otherwise,
I
can
keep
going
through
the
edges.
B
A
Okay,
can
you
can
you
send
that
one
to
me
here,
because
I
am
not-
I
am
not
actually
checking
slack
constantly
during
the
week.
A
But
our
qe,
apparently
it's
not
growing
that
fast.
I
can
see
some
things
in
helmshart
and
other
things,
but
so
I
think
it's
it's
just
a
matter
of
me
stopping,
maybe
and
taking
a
look
into
those
as
well.
A
A
Some
rumors
are
missed.
I
don't
look
fixing.
A
A
A
Okay,
this
one
is
closer
supporting
imported
blah,
blah
blah.
What's
this
one,
this
is
an
old
one
hold
on
I'm
just
gonna.
Take
a
look.
I
just
don't
wanna
miss
this.
A
B
B
A
It's
gonna
release
anyone
yeah,
but
that's
okay.
I
mean
carlos
keeps
the
track
on
the
release
things.
So
I
think
it's
for
this
one.
It's
fine.
A
Yeah,
the
bot
releases
automatically
every
time
you
bump
a
sharp
version.
This
is
why
there
are
some
discussions
if
we
should,
if
we
should
always
bump
or
not
because
for
the
users,
it
seems
that
we
are
constantly
releasing
new
home
charts,
but
I
think
it's
in
in
case
of
helm,
shards,
it's
fine.
We
are
not
actually
updating
the
main
in
chinax
version,
but
maybe
we
should
separate
versions
between
if
they
are
just
like
new
small
features
versus.
I
am
updating
the
ingress
controller
thing.
B
Carlos
asks
people
to
bump
help
chart
every
time
somebody
changes,
values
or
any
conflict.
The
problem
is
they?
Don't
I'm
not
certain
if
the
change
log
also
is
you
know,
updated
at
that
time,
so
I
think
either
we
should
tell
we
should
help
them
like
in
this
case
now.
The
change
log
has
to
get
one
more
line
right
for
this,
so
we
have
to
find
a
way
that
if
somebody
bumps
help
chart,
they
also.
B
A
Log,
the
problem
is,
if
you
update
change
log
in
in
the
rule.
Just
me,
james
and
and
jintao
can
approve
that,
while
in
this
case
can
be
approved
by
any
home,
chart
approver
but
yeah,
it's
it's
not
something
that
urgent,
but
we
need
to
take
care
of
that
yeah
yeah.
Okay.
This
is
done.
B
A
A
That
later,
just
just
make
sure
that,
because
this
is
going
to
get
merged,
then-
and
we
I
mean
I
can
hold
it,
but
then
I
would
just
like
this
one
to
to
get
measured.
So
if
you
can
just
later
just
change
the
change
log,
okay
got
it,
but
that's
not
gonna
be
published
right.
B
It's
not
so.
A
Okay,
so
let
me
just
finish
the
for
the
critical
ones
and.
B
In
the
short
time
we
have,
there
are
not
too
many
pr's
and
issues
that
no,
that
cannot
be
done
async,
but
what
I
think
it's
important
that
just
whatever
short
time
you
have.
We
need
to
sync
up
on
that
test.
Grid
issue.
I'll
just
tell
you
what's
happened
and
and
I'll
tell
you
what
I'm
going
to
do
next
and
you
need
to
just
tell
me
that
that's
the
way
we
want
to
go
or
we
want
to
go
a
different
way.
A
A
Okay,
can
you
can
you
just
maybe
close
some
some
other
of
those
and-
and
I
can
take
a
look
into
this-
just
take
a
look
into
the
issues.
If
you
do
have
still
something
that
needs
attention
from
us,
then
I
can.
I
can
go
through.
There
sounds
good.
A
Just
just
so
at
least
we
don't
lose
the
track
on
on
the
issues
and
we
can
go
to
the
test
grid.
Yeah
ingress
latency
appears
to
be
incorrect.
Some
of
those
they
are
pretty
old
cloud,
build.
A
A
A
This
is
this:
is
that
one
that
we've
been
discussing
on
on
adding
some
performance
tests
for
each
pr
right,
and
I
think
this
one
is
pretty
important
before
we
start
messing
with
other
things
like
the
gateway
api
and
so
on.
So
I
would
say
this
is
a
from
the
from
the
operational
side.
This
is
actually
a
priority,
at
least
for
me
right
so
yeah.
B
Okay,
okay,
so
I
was
looking
at
issues.
So
do
you
want
what
is
what's
the
plan.
A
B
A
See
yeah
yeah
a
simple:
we
should
have
like
a
simple
scenarios,
establish
those
simple
scenarios,
but
at
least
I
have
some
return.
We
we
shouldn't
be
consuming
that
much
cpu
and
memory
from
the
test
from
the
test
infrastructure
as
an
example,
but
at
least
doing
some
tests.
That
can
say:
okay,
this
is
gonna,
add
more
time
or
less
time
and
having
some
some
history.
So
this
is
this.
One
is
a
bit
important
for
me.
B
A
We
can
actually
have
some
help
from
graphene
folks
as
well,
but
my
idea
is
actually
taking
a
look
into
this
one,
just
just
to
make
sure
that
we
can
have
some
good
json
or
whatever
we
can
put
on
probabilities
later
and
take
a
look
into
that.
C
A
A
B
B
A
Okay,
let's
finish
the
important
ones,
so
at
least
we
may
have
no
happy
buildings
with
us
and
what's
this
one
fixed
monitoring.
A
A
A
A
Okay,
last
home
shard,
multiple,
okay,
this
is
the
goal
we
spoke
about
that
before
right,
so
I'm
gonna
keep
this
one
open,
sorry
and
I'm
planning
and
planning
actually
making
a
new
release.
Probably
this
weekend,
a
new
minor
release
fixing
adding
some
more
things
into
the
deep
inspector,
some
new
rules
there
and
I
can
actually
we
can
start
at
least
updating,
go
the
compilation
version
of
go
right
so
and.
C
A
B
A
A
Yeah,
which
we
don't
need
yet
because
314
is
actually
supported
right,
but
I,
as
I
said,
I
want
to
make
actually
a
major
release
when
we
update
to
315
updating
to
315
and
updating
open
rest
it
to
124,
which
has
been
developed
right.
So
as
soon
as
we
do
have
the
124
and
in
alpine
315,
then
I
want
to
make
a
major
update,
because
this
may
break
some
stuff
right.
So
I
would
like
to
have
a
beta
and
ask
people
to
to
test
the
beta
version
and
then
just
release
sounds
good.
A
We
need,
I
will
take
a
look
into
that
as
well,
because
updating
actually
ginkgo
to
to
version
two
is
a
breaking
change.
There
are
some
small
breaking
changes,
but
I'm
not
sure
why
this
is
hardcoded.
I
need
to
check
that
thing.
Actually,
let's
just
make
a
new
release
to
fix.
A
This
will
enter
this
new
okay,
and
let
me
just
take
a
look
here.
Okay,
do
you
want
to
update
go
or
do
you
want
me
to
update
go?
Do
you
want
to
make
a
try
on
that.
A
B
B
Will
open
the
facebook
page
and
go
to
click
on
today's
report
and
it
is
failed
yeah
and
if
you
go
to
the
end
open
the
hidden
files
122
last
one
which
one
there
are
some
files
and
line
lines
are
hidden
right,
so
we
open
the
hidden
lines,
yeah
some
deadlines,
so
basic
summary
is,
or
from
from
march
onwards,
the
the
error
was
ginkgo
not
found
right,
so
so
the
prop,
and
so
the
two
three
factors
one
factor
is
when
you
removed
the.
B
So
after
a
lot
of
these
debugging
commands
and
all
that,
we
figured
out
that
we
have
to
install
gingko
right,
it's
probably
an
image,
it's
a
so
basically
what
the
revelation
is
that
test
grid
runs
in
dind,
which
is
running
jenkins,
launching
all
of
this
in
kubernetes,
as
of
today
as
of
today
for
just
one
hour
30
minutes
ago,
one
hour
ago,
we
have
solved
the
problem
of
the
error,
which
is
ginkgo
not
found.
So
we
just
two
three
four
five
prs
finally
came
to
this.
B
We
installed
geeko
is
the
number
now
it's
still
failing,
and
there
is
simple
reason
for
that
is
that
this
is
the
ind
environment
and
it's
running
docker
script
right.
So
it's
launching
a
container
to
install
or
create
the
binary,
but
that
binary
is
getting
lost
in
that
container,
going
away
itself
right,
because
environment
is
the
pod
in
which
the
test
is
running.
B
So
that's
not
having
that
so
running,
docker
is
mounting
all
kinds
of
beautiful
things
required
right,
starting
from
the
docker
socket
to
the
volume
and
everything
so
whatever
gets
built
in
in
the
container.
In
this
case,
specifically
the
u2
test
binary.
It's
still
available
because
it's
built
and
copied
to
the
volume
right
in
this
case
it
is
not
copied
to
the
volume
it's
lost
with
the
container.
So
now
the
ginkgo
error
is
gone,
but
test
is
failing.
B
Still,
I
mean
overall,
test
is
failing,
so
if
so,
I
can
start
doing
more
debugging
like
where
which
environment
all
that,
but
basically
what
we
I'm.
The
two
approaches
is
either
with
wherever
we're
checking
the
dnd
enabled
we
again
set
it
to
false
right,
you
let
the,
if
condition
be
there,
but
set
the
di
and
d
enable
to
false
so
that,
instead
of
running
that
dind
command,
bash,
ginkgo,
etc,
it
will
use
the
same
run
in
docker
command.
B
A
Let
me
let
me
just
suggest
you
something
before
you
actually
start
doing
that,
which
is,
I
I'm
not
sure,
like
we
ever
checked
at
this
this
test
right.
So
one
thing
that
we
should
actually
ask
is
if
it
does
make
sense
to
us
to
have
this
test
there
right
anyway.
A
One
thing
that
you
can
do
and
I
cannot
prove
that-
is
actually
changing
the
type
of
the
job
from
periodic
to
let's
say,
pre-commit
or
post,
commit
or
something
like
that
and
adding
manual
or
something
like
that
right
and
then
we
can
trigger
that
thing.
You
can
you
can
open
a
simple
pr
and
we
can
we
can
trigger
these
manually.
So
instead
of
you
waiting
every
time,
this
thing
to
run,
you
can
just
do
a
slash
test,
periodic
something,
and
then
we
just
roll
back
and
it's
it's
fine
actually
doing
that
right.
A
The
other
yeah
yeah.
The
other
thing
that
I
was
going
to
suggest
you
is,
as
we
spoke,
don't
rely
on
the
don't
rely
on
that
on
that
image
that
that's
been
used.
Let
me
try
to
remember,
because
that
image
is
actually
maintained
by.
B
B
So,
even
if
you
put
our
image,
the
problem
is
still
going
to
be
the
same,
because
the
probe
the
see
the
specific
failure
is
the
build
of
the
e2e
test,
binary.
Okay,
so
even
if
it
uses
our
image
and
builds
that
binary,
it
will
still
remain
inside
that
container.
The
problem,
I
think,
I'm
facing
right
now
or
the
why
it's
failing
right
now
as
of
today,
is
that
it
builds
the
binary,
but
it
stays
inside
the
container,
we're
not
mounting
any
volumes.
A
Okay,
got
it
okay,
so
yeah,
so
instead
of
that,
just
just
try!
Okay,
you
want
to
disable
the
gene
here
right
is
this.
A
Yeah
try
this
try
changing
the
periodicity
here
from
12
hours,
maybe
removing
or
changing
this
from
periodic
to
something
manual,
so
ask
seek
testing.
How
can
you
do
that
right?
So
how
can
you
turn
this
thing
and
you
can
just
tell
them
like
you,
you
we,
we
decided
that
just
right
now,
so
we
don't
need
to
wait
every
12
hours.
You
can
change
this
to
something
else,
and
then
you
can
trigger
it
manually
doing
a
slash
test
with
the
name
of
this
test
right
and
it
it.
A
It
will
just
run
in
any
pr
right,
not
only
yours,
so
you
can.
You
can
use
any
pr,
that's
already
open
as
an
example
as
like
testing
this
this
this
job
here,
okay,.
B
A
Okay,
okay,
yeah
antonio
antonio,
is
great
he's
a
really
nice
person
so
yeah
just
and
if
you
need
something
just
just
ping
me
as
well,
we
we
can.
We
can
try
to
figure
this
out.
I'm
just
not
checking
that
much.
My
my
my
slack
in
kubernetes
workspace
but
yeah.
I
think
that's,
that's
a
good,
maybe
a
good
approach
as
you
don't
want
to
wait
for
the
periodic
one
to
run
every
12
hours,
just
change
this
to
a
pre-commit
or
a
manual
or
something
like
that.
A
Okay,
all
right,
so,
let's
see
okay,
I
am.
I
am
almost
on
my
my
fine
here
as
well.
So
aditya
carol
do.
Do
you
have
anything
else
you
want
to
discuss
for
today.
C
Not
from
me,
I've
been
a
little
busy
with
some
other
stuff
in
my
life,
so
I'll,
hopefully
get
back
in
a
few
weeks.
Okay,.
A
B
Yeah
there
is
some
important
ssl
stuff
being
reported,
and
it's
like
it's
not
it's
not
something
that
we
can
ignore
for
too
long.
So
when
you're
finished
with
the
high
priority
stuff,
we
need
to
kind
of
list
all
these
ssl
rated
issues
that
are
being
raised,
which
one
there
are
many
ss.
A
B
A
Yeah,
I'm
checking
there
are
some
folks
actually
opening
and
trying
the
phipps
one.
The
problem,
the
problem
with
fips
is,
is
that
how
can
I
say
themes
can.
B
Right
because
it's
not
even
possible,
he
agreed,
he
agreed
that
guy
agreed.
It's
not
fixed,
I'm
saying,
for
example,
right
now
there
is
a
eight
five
six
three
okay,
if
you
eat.
A
Okay,
I
I'm
not
I'm
not
against
that,
but
I
need
someone
actually
to
help
help
implementing
if
they
they,
if
they
can
help
on
that
more
than
welcome
them
to
to
review
the
pr.
I
just
I
just
can't
prioritize
this
one
right
now,
but
that's
that's
fine.
I
mean
we
can
accept
it.
If
this
is
something
acceptable,
we
can
accept
this
as
an
issue
and
maybe
do
a
help,
help
wanted
or
something
like
that.
B
Yeah,
it's
long
run,
but
I'm
saying
I
think
three
are
security
and
and
two
two
are
feature
related.
So
when
you're
finished
with
control
plane
and
all
these
priorities,
we
should
address
these
ones.
A
A
No,
no,
no,
no,
no!
No
yeah
yeah!
When
you
have.
Why
don't
you
have
some
time
I
will.
I
will
have
some
time
during
this
and
next
week.
Maybe
if
I'm
doing
let's
say
I'm
doing
nothing,
I
can
just
open.
I
I
keep
this.
This
spreadsheet
actually
open
in
my
machine.
So
I
will,
I
will
take
a
look.
I
will
keep
my
eye
on
that
and
maybe
look
into
the
ssl
ones:
okay,
okay,
okay,
yeah!
I
really
want
I.
A
I
really
want
to
reduce
our
kiwi
on
the
issues
and
and
and
prs
as
well,
while
that
I
I
have
I
I
I
I
found
this
one,
I'm
not
sure.
Okay,
you
are
on
this
one.
B
A
Yeah,
okay,
I've
missed
this
thing.
Yeah
demon
set
all
right.
Can
you
can
yeah?
Can
you
take
care
of
this
along?
Please
and
let
me
know:
yeah,
okay,
okay,.
A
A
A
Problem.
Okay,
so
I
think
that's
that's
all
for
sorry
folks,
for
for
today
I
will
be
away
next
week.
I
will
be
actually
around
so
yeah,
but
I
will
not
be
checking
as
much
my
my
slack
as
well
we'll
be
at
click
on
and
then
the
other
week
we
can.
We
can
talk
again
and
thank
you.
Thank
you.
Everybody
for
for
joining
us.
B
A
B
Yeah,
do
you
recall
from
from
memory
if
we
are
using
lin
anywhere
else
in
ci
or
anywhere
else,
we're
not
using
gene
right.
B
A
A
Probably
it's
something
legacy
probably.
This
is
why
I
told
you:
maybe
we
should
just
take
a
look
into
that
job
and
see
if
it
does
make
sense
to
us
actually
to
have
that
job,
because
we
don't
check
the
legacy
jobs
right.
We
don't.
We
don't
actually
keep
looking
into
that.
We
we
we
check
the
jobs
based
on
on
the
pr's
and
etcetera.
Should
we
be
checking
that
thing
first
right?
B
Got
it
so
I'm
sorry
taking
the
time
but
see
the
question?
No,
no,
no,
no
problem!
The
crux
of
the
matter
is
this:
okay,
when
I,
when
I
enable
that
when
I
put
a
echo
statement
to
check
the
value
of
enabled
or
not
when
I
run
make
e2e
test
on
laptop,
I
can
very
clearly
see
the
test.
Passing
hundred
percent
and
I
can
very
clearly
see
the
variable
showing
dind
is
not
enabled
okay.
So
that's
why
I'm
asking
accept
this
except
this
pro
job
in
in
test
grid.
I
it
doesn't
seem
like.
B
I
have
not
seen
any
I've
tried
everything.
I've
looked
at
ci
logs,
I've
looked
at
laptop
logs
and
I'm
looking
at
only
it
seems
on
test
grid.
Then
this
enabled
nowhere
else
is
being
used,
so
why
we
have
that,
if
condition
is
my
curiosity,
that's
why
I
asked.
A
I
have
no
yeah
okay,
I
have
no
idea.
I
have
no
idea
this.
This
may
be
legacy
as
a
as
I
told
you,
this
may
be
something
that
we
had
on
past
and
we
don't
don't
have
now.
Even
we
don't,
we
don't
run
our
I'm
checking
here.
We
don't
run
out
our
e2e
tests
as
an
example
in
in
pro
anymore
yeah.
We
run
it
on
github
actions
right
right,.
A
Maybe
maybe
one
thing
that
we
can
do
is
actually
drop
this
test
from
the
test
grid
and
run
a
periodic
job
on
on
github
actions,
because
github
actions
support
that,
and
then
we
just
run
our
periodic
test
there
and
send
some
alerts
some
email
or
something
like
that
right,
the
the
the
con
on
that
is
that
actually
we
are
not
going
to
have
ingress
in
ginax
as
part
of
the
test
grid.
But
to
be
honest,
we
are
not
checking
that
thing
right.
B
B
A
Take
a
look
into
the
gate,
maybe
maybe
I
did
maybe
not.
I
think
I
think
that
thing
was
already
there.
To
be
honest,
I
can't
remember
of
that
piece
of
code.
Actually,
okay,
okay,
okay,
so
I
think
that's
that's
all
for
today,
right,
yeah,
yeah!
Okay,
thank
you.
Folks,
take
care!
You
all
and
yeah
see
ya.