►
From YouTube: Kubernetes sig-aws 20180118
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
Good
morning
it
is
Friday
January,
12
2018,
and
this
is
the
sigit
of
West
meeting
our
first
together
this
meeting
of
2018
I'm
sure
we
have
a
lot
coming
this
year.
We
are
recording,
so
don't
say
anything
you
don't
want
on
the
record
and
otherwise
we'll
go
straight
into
the
agenda.
I
think
we
were
pretty
light
agenda
as
one
might
expect
for
the
start
of
the
year,
but
actually
only
item
we
have
on
it.
So
far
when
I
put
on,
which
is
just
to
be
in
case,
anyone
has
not
seen
the
meltdown
specter
bugs.
A
Please
be
aware
that
you
do
have
to
update,
or
it
is
highly
recommended
that
you
update
your
your
kernel
at
least,
and
there
are.
There
is
a
new
version
there.
For
example,
the
cops
kubernetes
maintained
a
mi,
but
you
know
every
every
distro
will
likely
have
an
update,
and
if
you
haven't
updated
this
basically
this
year,
then
it's
like
baby.
You
are
vulnerable,
I,
think
the
only
I
think
a
workaround
for
some
of
the
bugs
is
to
run
on
AWS
on
dedicated
tenancy
hardware,
which
will
help,
but
it
is
not
a
it
doesn't.
A
It
doesn't
protect
you
against
protest
process
and
you
know
I
think
we're,
although
still
freaking
out,
like
all
the
implications
of
it,
but
if
you
haven't
seen
that
definitely
update
your
kernel
as
soon
as
your
vendor
produces
a
a
an
updated
kernel,
an
updated
in
am
I
enough.
Anyone
else
wants
to
add
anything
about
any
other
mitigation
strategies
or
things
they
may
have
be
aware
of,
but
otherwise
yeah
just
make
one
make
double
sure
that
everyone
was
aware
of
that.
One.
A
B
Yeah,
so
today
is
the
last
day,
apparently
for
Cuba
and
CFP.
It
has
been
running
for
a
while.
So
the
keuken
CFP
ends
today.
So
if
anybody
is
interested
in
co
submitting
a
session
now,
I
am
submitting
a
few
sessions.
You
know
I've
been
coordinating
a
few
sessions
within
AWS,
but
if
anybody
is
interested
in
Co
submitting
a
session
or
submitting
a
session
of
yourselves,
make
sure
you
do
that
today
make
sure,
because
I
do
need
to
iterate
what
Q
Khan
is,
but
it
is
the
leading
event
and
the
attendance
has
been
growing
drastically.
B
You
know
from
Berlin
to
Aston
it
was
4x
growth.
The
numbers
are
gonna,
be
unprecedented
at
Copenhagen
and
then
in
Seattle
this
year
they
don't
even
know
they
don't
know
how
to
plan
for
it.
So
the
numbers
are
just
going
crazy,
so
I
would
recommend
all
the
good
work
that
we're
building.
We
will,
of
course,
have
the
CD
LW
s
face
to
face
I'm.
Sure
Justin
will
arrange
that.
A
B
And
I
guess
the
next
related
point
in
that
case
is
darker
Khan.
So,
even
though
darker
Khan
is
not
about
kubernetes,
but
the
recent
embracing
of
scuba
net
is
macao.
Kubernetes
can
be
easily
configured
in
dhaka
for
my
talk
or
ee.
So
if
there
is
any
sid
AWS
angles
that
you
guys
can
think
that
oh,
this
is
how
I
can
have
eks
cluster
configured
in
docker
for
EE,
for
example,
and
how
that
would
make
sense
or
a
cop
space
cluster
anything
well,
he
cares.
B
You
know
you
can
do
that
right
now,
because
it's
not
GA
but
a
cops
based
cluster
or
any
kind
of
cluster
running
on
AWS
I.
Think
that'll
be
a
great
angle
to
show
as
well
or
migrating
from
a
cops
based.
Cluster
I,
don't
know
from
mini
cube
to
a
coughs
based
cluster
using
some
darker,
tooling
and
I'm
just
throwing
out
ideas.
But
whatever
comes
to
your
mind,
you
know
the
talker
con,
CFP
I,
think
ends
Monday
or
Tuesday.
I.
Think
15
I
think
is
the
late
4-door
car.
B
A
That
would
be
it
certainly
with
with
Tucker
now
I
am
personally
tried
it,
but
effectively
I
think
they
released
the
version
now
where
there
is
communities
that
run
sort
of
almost
out
of
the
box
or
an
option
to
run
communities
out
of
the
box
with
doctor.
Is
everyone
installs
on
their
machine?
So
it
is
certainly
a
I
think
the
poridley
angle,
you
know
being
able
to
run
criminals
on
your
the
same,
manifests
on
your
machine
and
then
on
cops
and
ejs
and
gke
and
GCE,
and
maybe
even
Azure.
You
know
I
think
that
would
be.
A
A
C
Problem
is
there's
a
new
button
and
zoom
and
a
mute
button
on
my
iPhone
very
confusing
so
yeah,
that's
it
so
my
question
is:
is
the
responsibility
of
like?
Is
it
sig?
Aws
has
a
responsibility
to
take
care
of
the
spotting
or
is
it
named
follows
responsibility
to
take
care
of
this
bug,
I'm
leaning
towards
the
Installer,
because
until
the
instance
is
tagged
properly,
I
don't
think
it's
a
good
node
for
the
kubernetes
to
be
install
long,
so
I
think
I.
A
That's
certainly
something
we
that
is
on
the
table
like
previously
would
not
have
been
on
the
table
because
it
was
allowed
not
to
tag
things,
but
yes
it
that
if
we
don't
solve
it
in
the
cloud
provider,
the
problem
is
that
we
have
to
every.
We
need
every
installer
to
solve
it
right.
So
I'm
gonna
try
to
think.
Like
Greg
I
know
you
have
a
you
have
announced
Greg
Taylor
I
know
you
have
an
installer,
it
is
I
presume
you
would
rather
resolve
this
in
the
cubelet.
A
C
C
Some
people
could
comment
on
the
issue,
so
we
can
work
it
out,
I'm
kind
of
lean
more
towards
Installer,
because
you
know
I'm
not
sure
if
or
even
cool
it
knows
that
you
know
I
guess
coolers
can
do
the
ups.
The
call
to
get
the
the
base
URL
off
of
the
you
know
yes
and
since
fight
sometimes
clearly
doesn't
even
have
access
to
that.
So.
D
A
The
problem
is
just
the
tags
aren't
set
right,
so
it
could
just
it
could
just
wait.
Just
wait
for
those
tags
to
be
set
is
my
understanding,
but
which
is
what
what
a
exactly
what
something
that
came
before
cubelet
would
also
have
to
do
so,
like
cube
ATM,
for
what
in
pops
would
have
to
do
it
and
everyone
that
uses
anything
else
has
to
do
it
as
well,
but
presumably
only
if
they
use
the
consensus.
I
suspect
a
lot
of
people
will
just
put
him
asleep,
300
or
something
and
cross
their
fingers.
E
C
A
A
C
Yeah
I
think
so
Gerber.
It
has
been
saying
this
for
a
while.
You
comment
on
the
cops
issue,
so
this
has
been
around
for
a
while
I
use
finances
all
the
time,
but
there's
not
my
counts,
not
busy.
So
I
assume,
if
you
have
a
good
amount
of
node
turn
on
your
account.
That's
when
you're
seeing
this
I
don't
even
know
how
to
recreate
this.
Like
an
other
fun
partner.
It's.
A
Also,
it's
also
possible
that
happens
more
often,
and
we
just
see
it
as
flakiness,
for
example.
So
it
might
be
that
I'm
trying
to
think
of
like
exactly
what
happen,
but
if
it
may
be
that
just
weird
things
go
wrong,
that
we
attributed
to
general
flakiness.
That
is
not
fair.
To
tribute
to
flakiness
is
actually
this
business
shoe,
but
anyway,
yeah,
okay,
I
think
we
should
fix
it.
Thank
you
for
bringing
that
I
mean
that
went
up.
I
will
mark
it
as
serious,
I
guess
or
something
or
oh,
it's
pretty
serious.
A
C
So
I'm
wondering
some
people
that
are
hitting
this
are
hitting
a
bug
or
something
you
know,
there's
always
that
opportunities.
A
A
F
Was
me
Monty
here,
I
can
yeah,
as
I
was
looking
at
updating
our
cluster.
You
know,
because
of
Spectre
I
wasn't
clear
to
me
whether
the
batch
and
you
know
it
seems
like
brings
the
note
down
the
weights.
Then
it
brings
it
up
and
I
I
just
actually
don't
understand
exactly
how
the
bastion
is
working
and
whether
it
can
be
a
rolling
update
there
or
not.
So
I
was
just
curious,
so.
A
We
don't
so
currently
the
cops
passing
doesn't
do.
A
lot
is
just
a
machine
that
runs
the
same
image.
It's
it's
in
in
tone,
it's
in
its
own,
auto
scaling
group,
and
it
actually
doesn't
run
anything
and
one
of
the
things
we've
fought
it
like
it
doesn't
even
run
cubelet.
One
of
the
things
I
thought
in
the
past
would
be
cool.
A
You
know
to
have
it
actually
be
a
cooper
Nettie's
node
that
actually,
you
know,
was
able
to
run
like
in
general
like
it
would
be
not
of
just
cops
like
there
was
able
to
run
like
a
proxy
or
was
able
to
run
OpenVPN
or
was
able
to
run
nothing
if
you
just
wanted
to
ssh
in
across,
but
like
the
downside
of
that
would
beat
it.
There
will
be
a
little
more
attack,
surface
area
because
it
would
be
running
cubelets,
so
it's
a
little
different
from
the
bastion
so
as
cops
implements
it
today.
A
It
is
just
a
node
that
it
runs.
That
runs
just
the
image
and
doesn't
actually
run
anything
else,
but
your
SSH
key
is
on
there
goodbye
the
normal
AWS
mechanism,
Chris
I
guess
you
can
I
mean.
There's
always
there's
always
so,
given
that
you
can
always
just
like
scale
up
the
scale
up
the
auto
scaling
group,
and
you
know
that
sort
of
thing
but
Chris
you
wrote
the
rolling
update
here.
I,
don't
know
if
you
were,
if
you
remember
off
the
top
of
your
head,
how
it
works
with
the
bastion
well
I.
C
C
You
wouldn't
necessarily
lose
connectivity,
but
you're
not
gonna,
be
able
to
guarantee
when
the
bastion
will
be
going
up
or
down
I'm,
not
certain
what
you're
trying
to
prevent.
Are
you
trying
to
prevent
losing
all
access
to
the
cluster?
You
have
something
running
through
the
bastion
I
want
to.
Why
don't
we
do
this
one?
Why
don't
you
ping
me
after
the
call,
okay,
Sherlock?
Okay,
thanks,
we
can
get
no
word.
Okay
and.
A
F
A
A
For
like
it's
like
a
backdoor,
if
you
don't
have
a
direct
VPN
connection
or
another
mechanism
into
your
V,
PC,
alright
see
ya
when
you
sorry
for
running
when
you
have
no
public
eye
piece
right,
yeah!
Well,
okay,
that's
great!
Okay!
Thanks
for
clearing
that
up.
If
there's
anything
else
on
the
agenda,
all
of
these
anything
else,
anyone
wants
to
talk
about
yeah.
A
Although
idea
that
might
be
a
shorter
meeting.
I
imagine
everyone
is
busy
hitting
refresh
on
kernel.org
in
there
am
I
provider
or
there
distros
yeah
I,
think
that
my
sense
of
that
is
that
we're
gonna
see
basically
weekly
updates
for
the
next
couple
of
months.
So
that's
gonna
be
fun,
but
I'll
see
how
bad
it
gets.
I.
D
Do
have
one
quick
thing
that
not
maybe
not
so
much
an
agenda
item
but
I
I
filed
a
PR
on
the
the
doc
site,
the
community
stocks
for
annotating,
all
the
EOB
annotations
for
classic
al
beefs
and
I.
Did
it
just
before
the
holiday,
so
I
think
that
kind
of
got
lost
in
the
like
holiday,
New,
Year,
stuff
I
can
put
a
link
to
it.
I
just
need
so
that's
a
WS
folks
or
community
folks,
but
to
just
have
a
have
a
review
on
that
a
technical
review.
A
Guess
we
couldn't
ask,
is:
is
there
anything
else
anyone
is
working
on
in
the
800s
space
they
want
to
share
or
think
about
working
on,
like
any
New
Year's
resolutions
other
than
the
eks
thing.
I
guess
some
of
you
are
working
on
so.
B
What
are
the
fun
ideas
that
was
coming
up
in
a
meeting
I
think
it'll
be
fun.
To
implement
now
have
a
custom
resource
definition
that
could
take
ECS
tasks
and
deploy
it
as
a
kubernetes
resource
I
mean
that's,
not
a
WS
specific,
but
that's
in
general.
It
is
taking
an
AWS
technology
and
deploying
it
in
a
cop's
cluster,
for
example,.
B
It'll
be
fun
one,
you
know
if
anybody
is
interested
in
working
on
that
you
know
we
can
create
something
like
that
together
and
and
I've
been
reading
through
the
CRD
documentation.
Is
there
but
sort
of
not
there?
You
know
for
somebody
who
wants
to
write
a
CR
d
by
itself.
It's
not
very
easy
and
straightforward,
so
I
think
that'll
be
a
cool
tutorial
that
we
could
write
and
I
would
be
happy
to
take
that
as
a
contribution
on
our
open
source
blog.