►
From YouTube: kubernetes sig-aws 20190628
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
Hello,
everybody:
it
is
Friday
at
June
28th.
This
is
sick,
AWS
I,
am
your
moderator,
facilitator,
Justin,
Santa,
Barbara
I
work
at
Google,
a
reminder.
This
meeting
is
being
recorded
and
will
be
put
on
the
internet
and
therefore
fitting
mindful,
as
always
have
our
communities
could
have
conduct,
which
can
basically
be
a
good
person.
A
Today
on
our
agenda,
we
J
at
our
last
meeting
proposed
that
we
go
through
basically
116
grooming,
where
we
look
at
our
github
issue.
Backlog,
for
example,
look
at
what
the
things
are
that
that
we
should
probably
look
at
in
116,
decide
whether
we
can
achieve
them
in
116
on
session
the
PR
front,
but
basically
like
sort
of
try
to
show
that
process
and
go
through
that
more
collaboratively,
often
the
github
experience
can
be
confused
about
trees.
A
So
if
we
do
it
together
that
and
put
it
on
recording,
then
hopefully
everyone
can
see
it
and
understand
our
collective
thought
process.
I'm
gonna
try
to
share
my
screen
now
before
we
do.
That.
Is
there
anything
that
anyone
else
would
like
to
bring
up
that,
if
not
related
to
a
1/16
grooming
type
activities.
A
Apologies
for
the
delay
there.
We
are
come
on
top
okay,
so
we
should
see
I
believe
the
prowl
command
help
is
like
what
people
see
yeah.
Thank
you,
okay
get
alright,
so
we
might
not
always
do
this,
but
because
I
should
have
permissions
to
apply
labels
and
things
directly,
but
in
general
this
is
proud.
A
Let's
see
so
most
people
can
apply
labels.
So
if
we
want
to
add
an
area
from
very
proud,
but
basically
what
we
add
area
AWS
a
lot
or
even
remove
area,
eight
of
us
a
lot
and
what
those
correspond
to
sorry.
So
those
are
its
this
command
here
and
these
correspond
to
the
labels
and
the
labels
that
we
have
open
are,
let's
pick
a
PR
that
might
have
something
to
do
with
it
of
us
to
go
into
or
yeah.
A
A
You
see,
there's
a
bunch
of
labels
here
and
I'm,
not
entirely
sure
which
labels
we
actually
want
to
use
today,
but
I
think
probably
we
should
use
area.
We
should
use
the
gate
address.
So
if
we
say,
for
example,
this
one
already
has
the
Szigeti
bus
label
so
and
looks
correct.
Let's
try
to
find
one
which
does
not
and
let's
just
add
that
label
I.
A
Okay,
so
here
we
are
Camus,
that's
Trammell,
which
is
a
awsm
which
is
not
labeled.
Let's.
A
About
this,
that's
EE,
alright,
perfect.
So
this
one
is
a
an
issue:
support
UDP
with
a
to
us
and
I'll,
be
kind
bug
needs
sick,
and
so,
if
we
go
into
this
and
we
comment
on
it,
we
should
be
able
to
write
I
believe
sig
AWS.
Let
me
just
confirm
yes,
sig
testing,
so
sig
AWS
and
I
as
a
anyone
user
I'm
able
to
come
with
that
I
think
you
might
have
to
remember
of
the
kubernetes
org,
but
there
we
are-
and
we
have
this
to
get
this
okay
great.
A
So
that
is
the
first
step
in
our
triage,
which
is
basically
to
go
through
and
make
sure
that
everything
is
tagged
appropriately
and
so,
for
example,
let's
see
just
I,
don't
know
if
we
should
tag
a
few
more
like
that,
didn't
work
there
we
go
I,
don't
we
should
take
a
few
more
and
then
we
can
just
focus
then
on
the
ones
which
are
tagged
with
city
AWS,
I'm,
just
gonna
go
through
the
recent
ones.
I
don't
see
any
it
looks
like
someone
may
have
been
going
in
and
fixing
them
for
me.
Thank
you.
A
A
A
Yes,
I
guess
the
answer
here.
This
is
really
yeah.
Next,
that
is
not
our
it's
not
technically
our
issue.
It's
not
specific
to
us
and
I
guess
I
needed
us.
We
also
could
use.
Then
I'll
be
right
anyway.
Moving
on
all
right.
So
now
we
can
do
our
main
triage,
which
is
open
issues
open,
PRS,
labeled,
the
sigit
of
yes,
and
we
can
assume
that
basically,
these
are
more
or
less
good
and
still
do
the
same
thing,
but
do
the
same
thing
for.
A
Okay,
so
let's
look
at
our
poll
requests.
First,
I
guess.
The
important
thing
is
to
like
look
at
these
and
understand
whether
any
of
these
are
good,
so
they're
about
12
that
are
open,
so
I
guess
we
should
figure
out
which
ones
we
definitely
want
to
achieve
in
116
and
the
cumin
area
is
116
deadline
or
release.
C
A
A
Thank
you
for
thank
you
for
finding
it.
Alright,
there
is
look
like
a
lot
of
these.
We
basically
should
lots
of
cherry
pick
a
lot
of
these.
We
should
basically
get
in
I'm
trying
to
go.
There's
anything.
We
would
not
want
to
get
into
116,
but
let's
put
these
into
the
milestone
and
then
we
can
have
a
look
at
it.
A
A
Okay,
all
right,
so
we
have
put
our
first
one
into
the
milestone.
I,
don't
know
whether
anyone
else
can
do
that,
but
I
feel
like
this
going
through.
These
feels
less
important,
and
that's
anyone
wants
to
call
attention
to
any
of
these
PRS.
It
feels
like
we
definitely
have
a
look
at
all
of
these.
Some
of
the
older
ones
are
basically
pretty
controversial.
C
Very
oldest
one
there
that
annotation.
If,
if
we
have
an
annotation,
you
have
to
probably
means
it
that
one
requires
just
some
additional
testing.
The
patch
itself
looks
fine
I'm,
just
a
little
suspicious
of
it,
not
not
not
because
there's
anything
Shady
just
because
I
don't
understand
why
the
author
thinks
it
works,
but
if
it
does
work
correctly,
it's
a
good
candidate
for
1/16,
okay,.
A
Let's
see
so
all
right
well,
leave
that
open
and
it
looks
like
I,
don't
know
if
it's
been
revised
since
yeah
I
think
I
had
a
look
at
this
and
I
also
had
some
questions
with
you,
like
I,
showed
the
same
questions
about
how
it
works,
so
I
think
we
can
probably
talk
about
getting
it
in,
but
should
we
tag
it
for
the
milestone
or
it
feels
weird
to
check
if
the
milestone,
if
we
can
at
least
understand
it,
I
guess
have
a
bet.
I
would.
C
A
And
I
it
is
unstable.
It
looks
like
and
ie
well
you're
already
assigned
to
it.
So
I
think
it
looks
like
we
can
yeah.
Okay,
that's
that's
a
good
one!
Yeah
I
should
for
160
minus
1.
Thank
you
for
looking
at
that.
Alright!
So
those
those
are
some
of
the
I,
don't
wanna.
So
it's
calling
attention
to
any
other
PRS.
That
is
not.
A
A
That
is
expensive
and
often
we're
querying
by
name
just
to
look
up
the
ID,
the
instigators
instance
ID.
Often
we
go
from
a
node
to
the
node
name
to
the
instance
ID,
but
the
instance
ID
was
actually
just
in
the
node
object
anyway
in
the
provider
ID
or
external
ID,
or
wherever
we
put
it.
So
yes,
I
think
it
would
be
great
to
optimize
that
there's.
C
Two
there's
two
approaches
here:
I
know
Justin.
Several
months
ago,
you
proposed
a
fairly
aggressive,
but
definitely
correct
approach
to
fixing
this.
This
PR
that
I've
been
following
is
a
little
bit
more
of
a
kind
of
tactical
fix
to
get
us
through
some
some
weight
limit
issues.
So
I
think
we
should
push
this
one
forward,
but
also
take
another
look
at
kind
of
your,
your
bigger,
better
fix
the.
A
A
I
feel
like
yeah:
let's
talk
about
that
I
guess
we
should
talk
about
that
now
in
sig
cloud
provider,
it's
sort
of
a
cross-cutting
thing.
If
I
recall
correctly,
though
the
real
problematic
one
was
the
volume
in
interface.
So
we
should
what
you
guys.
We
should
have
a
look
at
that
as
well,
but
yeah
I,
I,
guess
I'll
have
another
look
at
that
as
well
and
see
if
I
can
where's
my
key
I
wonder
if
one
if
we
got
closed.
C
A
Yeah
that'd
be
great,
I.
Think
I
think
that
would
be
a
good
thing
to
happen
in
that
in
that
group
and
also,
if
you
have
that
PR
to
add
the
SE
queda
vs
label
to
it,
because
it
should
be
on
this
list.
I
don't
see
it.
So
obviously
we
we
are
missing
some
things,
all
right,
so
I
think
in
terms
of
the
poll,
requests,
I,
think
I.
Think
people
get
the
idea.
A
D
D
A
A
A
All
right
so
I
guess
that
one
IO
guess
aid
of
us
just
added
support
for
UDP
in
NLB.
If
that's
right,
so
that
feel
makes
sense
that
that
would
be
a
new
one.
Probably
should
be
a
feature
request.
We
can
track
it
as
such
yeah
labels,
I'm
gonna,
look
at
the
name
of
the
label
and
I'm
gonna
see
so
it's
kind
feature.
So
I
could
click
there,
because
I
have
permissions,
but
I
will
try
the
one
which
I
believe
everyone
can
do.
A
F
A
B
G
A
Okay,
so
it's
particular
teams
that
can
do
it.
So
alright,
that's
disappointing,
but
alright
I
will
issue
the
command
and
we've
learned
that
you
have
to
be
in
order
get
into
the
milestone
has
to
be,
and
we
can
actually
mark
this
one,
as
this
would
be
a
great
help,
wanted
I
think,
possibly
even
a
good
first
issue,
because
I
don't
think
it's
terribly
difficult.
A
Hopefully,
I
mean
the
problem
with
all
of
these
is
they
seem
easy
and
then
we
get
into
the
weeds
and
there's
some
issue,
but
there
we
are.
Let's
see
if
that
works.
I,
think
that
is
probably
a
reasonable
first
issue,
based
on
my
anticipation
of
how
it
works
and
if
L
must
cycle
that
and
discovers.
It
is
not
then
feel
free
to
contact
me,
but
I
hope
it
is
not
too
bad.
A
C
A
B
F
A
A
C
Okay,
there's
a
whole
bunch
of
issues
related
to
the
aquarium
by
instanceid
I've
kind
of
got
some
visibility
into
a
lot
of
them.
I
just
apparently
am
inadequate
when
it
comes
to
adding
things
to
the
milestone,
so
maybe
I
should
just
tag
those
all
off
and
like
send
them
to
you
in
an
email.
Even
mass
update,
something
yeah.
A
A
C
A
You
that
would
be
wonderful
and
I
think
yeah,
but
I
think
we've
we've.
We
definitely
have
discussed
already
the
way
in
which
people
can
can
contribute
and
can
help
and,
like
I,
think
the
process.
I
guess
the
the
remaining
one
way
to
spend.
The
remaining
time
we
have
today
is
to
go
through
just
all
the
issues
and
try
to
work
on
them
together
like
triaging
them
all
or
oops
or
or
we
can
do
other
things
that
people
would
like
to
draw
attention
to
anything
or
would
like
to
spend
the
time
some
other
way.
I.
A
D
A
D
C
A
Yeah,
what
I
suggest
is:
if
no
one
else
has
anything
they
would
they
want
to
discuss
or
wants
a
particular
draw
attention
to
or
bump
to
the
top
of
the
heap
we
can
go
through
them
together.
Anyone
that
wants
to
drop
and
get
half
an
hour
back
is
welcome
to
drop.
Otherwise
we
can
sort
of
go
through
them
sort
of
in
order
and
spend
that
time
and
it
will
be
recorded
and
we'll
have
that
going.
Does
that
sound
good
to
people
or
does
that
I
guess?
Is
anyone
object,
I.
H
Issue
this
week
that
I'd
like
to
bring
up
as
I,
updated
to
two
communities,
one
thirteen
and
it
had
like
that
I
guess
it's
kind
of
like
a
breaking
change
because
it
stopped
might
McCluster
stop
being
able
to
communicate
with
ECR
and
I
needed
to
add
a
new
security
group
in
order
to
do
that,
I'm,
not
sure
if
anyone
else
ever
run
into
that
yet
I
know
there
was
an
update
to
the
SDK
m113
for
communicating
to
ECR.
Wait,
try
the
link.
Did
you
open
an
eight-oh?
Is
it
privately?
H
A
A
I
This
one
there's
this
isn't
like
a
there's
also.
There
is
also
the
issue
of
and
I,
don't
remember
how
far,
but
we
back
ported
this
issue,
but
when
ECR
introduced
private
link,
they
changed
the
API
endpoint
that
private
link
resolves
to
for
ECR,
so
that
SDK
change
had
to
make
its
way
backwards
in
kubernetes
and
I
should
think
it's
only
in
like
115
and
maybe
114
I,
don't
think
it
got
to
113.
So
if
you're
gonna
try
to
use
present
link
with
ECR
I,
don't
think
it's
going
to
work.
I
I'd
have
to
double-check
on
the
on
how
far
that
fact,
I
got
traffic
but
I
think
it
didn't
even
make
113.
Oh!
Is
it
just
in
eks
when
it's
justin,
we
pat
yeah,
we
patch
the,
are
cubelet
for
113
to
include
that.
Should
that
fix
it's
just
it's
an
updated
aw
sdk,
as
always
yeah,
so
sailing
upstream
113
cubelet
will
not
support
private
link.
Ecr
got.
H
A
H
I
H
A
H
A
I
would
suggest
it
sounds
like
there's
enough
here
in
terms
of
like
possibly
missing,
release
notes.
It
might
be
something
in
the
eks
bill,
but
it
sounds
like
there's
something
in
open
source
as
well,
that
we
might
know
documented
right,
mica,
so
I
guess.
If,
except
that
you're
happy
to
open
an
issue,
then
we
can
figure
it
out
from
there
sure.
A
A
C
A
C
A
A
I
wouldn't
have
to
yeah
we'll
have
to
follow
up
on
that.
There
is
direct.
There
is
direct
chat
and
zoom.
If
you
want
to
like
you,
can
hear
it
appear
with
each
other,
as
it
were
all
right,
thanks,
yeah
I,
wouldn't
I
don't
know
that
consequence
of
putting
those
in
public
stuff
so
I
guess
that
was
great.
Is
there
anything
else
any
any
other,
similar
preemptions,
I,
guess
of
high-priority
items
that
people
would
like
to
bring
up?
A
A
A
A
A
That's
definitely
one
that
I
can
at
least
start
up
yet,
and
it's
relating
to
caching
as
well.
It's
a
little
different
I
think
all
right
Seth
is
that
you,
by
the
way,
no.
A
It's
a
different
different
set,
all
right:
okay
and
the
other
one.
We
had
ice
it
back
to
the
list.
A
B
A
A
B
A
E
A
Isn't
a
we
do
create
a
single
security
group,
but
the
security
group
themselves
have
rules
underneath
them
and
we
aren't
coalescing
them.
Yeah
I!
Think
that's
because
this
person,
I
presume,
is
working
around
the
communities.
The
back
community
services.
Don't
have
port
Rangers
by
like
doing
port
80
81
82,
83,
84
85.
We
then
create
six
rules,
security,
group
rules
and
the
ask
would
be
that
we
do
combine
them.
B
A
Okay,
I,
don't
think
this
can
therefore
yet
to
go
into.
It
is
a
is
tag,
time
feature
which
is
correct:
I,
don't
think
we
can
put
into
a
milestone.
We
don't
really
understand
it
enough,
yet
it
certainly
sounds
non-trivial
as
well,
but
it's
not
totally
trivial
and
I
particularly
worry
about
what
happens.
If
we
there
can
be
some
weird
edge
scenario.
Cases
like
you
set
up
these
things
and
then
you
remove
one
from
the
middle
and
then
we
hit
another
limit
or
something
or
we
accidentally
start
deleting
other
ones.
That
I
don't
know.
A
A
I
I
I
A
different
one
that
I
can
find
her
I,
don't
think
that
one
is
on
I,
don't
know
that
that
one's
on
github,
yet
okay.
A
A
A
Interesting
note
kasha:
here
we
go
my
kiss
there's
the
thing
it
is
indeed
a
we
could.
Probably
this
is
the
bigger
one.
So,
let's.
A
We
can
we
can
cross-reference
them
to
each
other,
so
if
we
could
use
this
as
the
tracking
issue
for
the
whole
like,
oh,
this
is
okay.
Well
anyway.
Yes,
we
can
cross-reference
them
to
each
other
when
we
see
more
ones
like
this
seven,
eight
four,
oh
nine,
let's
go
I
can
remember
that,
and
it
is
indeed
a
feature
secret
of
us,
a
cloud
provider,
six
scalability
that
looks
great,
there's
no
milestone,
which
seems
correct.
It
isn't
something
that
we
we
have
any
real
way
of,
targeting
to
particular
milestone
and
yeah
I.
C
A
C
A
A
B
B
A
A
A
A
F
A
C
The
long
term,
yes
in
the
very
short
term
this
year,
hopefully
this
isn't
scoped
for
the
out
of
tree
migration,
but
what
we've
been
doing
so
far
is
just
putting
fixes
in
tree
with
the
hope,
I'm
just
kind
of
pulling
everything
out
all
at
once.
So
I
think
it
makes
sense
to
fix
this
entry,
knowing
that
it's
going
to
go
out
of
tree
yeah.
A
C
A
So
I
think
that
that
is
a
so
there's
an
answer
that
your
question
I,
don't
know
how
I
work
is
that
right,
yeah,
yeah,
yeah
Isis,
starts
in
tree
in
staging
at
some
stage
the
we
pull
the
cord
and
cut
the
cord
and
we
move
it
to
the
other
repo
which
I
believe
it
already
exists.
Right.
The
kubernetes
slash
ADA
mist
cloud
provider
where
it's.
C
C
A
Think
that'd
be
great
if
you've
got
the
agenda
open
to
have
a
great
thing
to
do
like
the
you
as
I'm
sure
you
know,
there
are
BOTS
which
can
copy
stuff
between
staging
and
other
repos,
so
we
may
be
able
to
mirror
into
here
and
then
reconcile
in
a
sort
of
controlled
way
and
actually
see
the
disks
right
because
is
gonna,
be
any
discs.
You
have.
C
Yeah
I
think
that's
a
good
thing
to
do
at
this
point.
We
kind
of
abandoned
the
one
approach
and
are
now
kind
of
heading
towards
maybe
going
back
to
the
mirroring
approach.
So
I'll
answer
that
honestly
cloud
provider
agenda
as
well,
and
we
can
kind
of
talk
through
that
there
because
I
think
that's
more
broadly
applicable
to
other
cloud
providers
as
well.
I.
A
Agreed
yeah,
that's
great,
and
thank
you
for
bringing
that
up.
Everyone
I
think
the
four
people
don't
know
it's.
A
cloud
provider
meets
on
sea
Wednesday's
of
this
week,
bi-weekly
at
1
p.m.
Pacific,
maybe
but
I
could
well
be
wrong
on
that,
but
it's
around
then
anyway
so
and
they
did
not
have
a
meeting
this
week.
The
next
meeting
should
therefore
be
in
12
days.
I
would
think.
B
A
A
A
He's
reporter
said
a
segfault,
there
is
indeed
a
segfault,
so
a
segfault
is
always
a
bug.
It
needs
to
be
fixed
in
my
mind,
so
this
has
to
go
into
the
next
milestone,
regardless
of
what
actually
happens,
but
it
should
never
be
a
segfault.
I
will
definitely
have
a
look
at
this.
Another
nice
ones
have
a
look
but
all
right
and
then
the
actual
problem
is
they
failed
to
get
the
logs.
Maybe
they
fail
to
connect,
looks
a
bit
really
failed
to
connect
to
the
actual
pod.
I
would
guess.
A
Okay,
anyway,
that's
looking
alright
next
on
the
list,
so
we
put
that
in
the
milestone,
because
any
signal
should
be
a
fixed
clamp
provider
anywhere,
should
query
instance
by
ID
again
have
I
somehow
next,
okay
anyway,
similar
to
that
one.
Let's
come
back
to
that
one
Atos
ipv6
route
cause
destination.
Sidewalk
is
invalid.
I
think
the
answer
is
yes.
It
does
because
we
don't
get
square
at
maybe
six
everywhere,
but
it's
coming
I,
don't
know.
If
there's
really
anything,
we
can
do.
F
A
There
is
an
open
issue
for
our
open,
PR
I
should
say
or
yeah
dual
stack,
which
should
come
in
soon
or
should
make
it
into
116
our
guests
and
I.
Think
actually
I,
don't
know.
If
someone
correct
me,
I
think
you
guys
might
work
if
you
just
have
a
pure
ipv6
network
today,
or
at
least
at
115
I
should
say
I
don't
know.
If
anyone
is
that
modern
or
that
futuristic.
A
A
A
A
A
So
all
PRS
go
into
the
master
branch
so,
as
you
say,
that
will
not
make
it
out
until
116.
If
you
do
want
to,
if
you
are
a
if
you're
the
author
appear
or
even
if
you
just,
if
you
just
see
a
PR
that
you
would
like,
you
can
do
a
cherry
day,
so
anyone
can
propose
cherry
picks.
There
is
this
script
a
hack
and
cherry-pick
pull
the
Sh
which
usage
instructions
there.
We
are.
A
A
I
guess
who
will
sort
of
make
sure
that
the
cherry-pick
is
sort
of
worthwhile
so,
like
there's
a
risk
with
everything,
so
is
it
worth
the
risk
and
is
it
fixing
a
bug
or
is
it
trying
to
sort
of
get
a
new
feature
in
and
typically,
if
it's
trying
to
introduce
a
new
feature,
the
cherry-pick
will
not
be
allowed
it.
We
sort
of
be
rejected,
typically
the
tripi
rejected
and
then
the.
A
But
if
it's
a
regression
or
a
bug
fix,
it
should
be
allowed
as
long
as
it
isn't
too
disruptive
sort
of,
in
other
words,
if
it's
proportionate.
So
if
it's
a
if
it's
a
one-in-a-billion
bug
that
like
has
a
pretty
high
chance
of
blowing
up
the
release
that
wouldn't
be
allowed,
but
if
it's
a
pretty
small,
sensible
fix
that
is
proportional,
then
it
would
be,
and
you
do
not
have
to
be
the
author
to
propose
a
cherry-pick.
A
So
if
you
see
a
fix
going
into
master
that
you
would
like
to
cherry-pick
I
believe
the
etiquette
is
to
say
you
know
I'd
like
to
see
this
in
1:14
and
give
someone
a
chance
to
reply,
and
then
you
can
run
the
cherry-pick
script
to
propose
that
cherry-pick
yeah
thanks
cool,
that's
that's
good
I
feel
like
we.
We
touched
on
a
bunch
of
other
stuff
along
the
way
as
well.
We
are
at
time,
I,
don't
know.