►
From YouTube: Kubernetes AWS Infrastructure - 20230223
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
This
meeting
is
being
recorded,
hi
Welcome
to
the
meeting
about
the
new
occupancy
infrastructure.
Just
remember
that
these
meetings
is
recorded
and
going
to
be
published
after
on
YouTube.
So
I,
as
you
know,
I
highly
suggest
to
be
mindful
so
everyone
in
this
call.
So
let's
start
we
basically
for
anyone
coming
to
anyone
coming
to
these
meetings,
There's
the
link
of
the
dogs,
the
zoom
chat.
C
A
That's
nice,
okay,
so
AR
review.
Let
me
share
my
screen.
First
Air
review,
so
people
Mighty
folks
have
access
to
an
account
for
pro
experimentation.
What's
the
update
on
that
problem.
E
Yeah,
so
Marco
could
not
join
because
we
are
currently
doing
a
release
on
our
side,
so
he
created
he
created
the
cluster.
He
created
it
with
a
terraform
and
with
the
terraform
manifest
they
are
in
place
he
had
to
after
he
talked
to
you,
I
think
I
know
he
had
to
destroy
the
Clusters
again,
because
we
need
to
recreate
it
with
kubernetes
1.23
instead
of
1.24,
because
Pro
got
some
issues
there.
E
A
Yeah
I
think
we
need
to
have
a
pull
request
of
the
telephone
code,
yeah
for
review,
yeah
and
yeah,
so
basically,
basically
just
see
what's
what's
missing
and
basically
outside.
What's
missing,
because
my
initial
review
I
saw
a
lot
of
things
missing
in
the
first
draft,
so
I
think.
Maybe
other
people
are
interested
to
review
that.
E
A
B
B
A
I
I
I
saw
that
the
bit
the
basically
the
idea
is
like
not
only
me
saying
seeing
this
other
people.
That's
why?
Basically
I'm
asking
cultural
press
like.
E
Yeah,
but
he
wanted
to
update
it
to
1.23
so
that
everything
is
in
in
in
the
correct
way
before
it
opens
the
pr.
A
Ok,
okay,
so
okay,
next
AI
for
next
meetings
have
a
case
closer
up
and
running.
E
B
F
B
Yep
point
of
just
clarification:
next
meeting
is
Sig
Kate's
infra
meeting,
not
AWS
infra
meeting
correct.
A
A
Okay,
next
day,
I,
basically
I
asked
for
the
costs
on
AWS
for
last
year.
I
got
that
information.
I
will
put
that
in
the
annual
report
for
2020
so
foreign
any
question
about
this.
B
B
Yay,
we'll
actually
be
able
to
see
as
we
shift
things
over
and
as
we
like.
If
we
move,
if
in
well,
fastly
is
a
bad
example,
because
that's
the
internal
spend
but
like
as
we
reduce
costs
it'll
be
a
little
bit
easier
to
show
and
as
we
shift
things
over
to
AWS
I'll
actually
be
able
to
meaningfully
show
hey,
we,
you
know
shifted
10
over
to
AWS.
A
That's
you
trigger
one
question
then
comment
so
I'm
gonna
I'm
gonna
go
to
the
next
one.
Yep
next
item
is
basically
Identity
or
federal
access.
Mermaid
is
working
on
the
PLC
with
Azure.
So
that's
trigger
a
question
about
odd
zero
and
OCTA.
Where
are
we?
Do
we
have
an
update.
A
A
G
A
A
With
GitHub,
because
GitHub
is
for
us
the
another
person
project,
the
only
way
where
we
have
a
centralized
platform
when
we
are
all
the
contributors,
so
the
idea
is
to
do
it
basically
enable
a
way
to
authenticate
to
AWS
on
for
specific
people.
Now
the
multiple
option
on
the
table.
My
bed
is
working
on,
basically
Azure
Plus,
Azure,
ID,
I.
Think
there's
a
POC
somewhere
there's
a
pool
request
somewhere.
It
be
tagged
me
on
it.
A
G
I
would
say:
GV
try
to
get
the
license.
Yep.
C
A
That
was
my
admit
and
recently
post
about
basically
he
posts
about
each
epos
I
think
20
minutes
ago
about
trying
to
do
that
and
it's
not
working.
So
maybe
we
can
like
say
open
source
options,
trying
to
look
at
tools
like
key
clock.
A
What
I
was
thinking
Dex
anyone
anyone
interests,
work
on
that
feel
free
to
basically
do
a
POC
and
tell
us
how
this
is
working.
That's
like
oh
bro
from
scratch,
something
in
trouble
and
basically
says
something:
Taylor
Made
for
us
that
only
I,
love,
git
and
plug
AWS,
and
maybe
Google
workspace
I.
Don't.
A
Now
mama
is
looking
at
Azure
ID.
If
that's
worked,
we
can
maybe
talk
to
Microsoft
and
see
how
we
can
get
the
Azure
idp1
licenses
and
we
basically
solve
the
problem.
Oh
GPU
might
come
and
tell
us.
Cncf
is
happy
to
pay
the
OCTA
subscription
because
that's
the
official
identity
provider
support
by
AWS
SSO.
A
We
also
saw
the
problem
Oh,
someone
come
and
say:
oh
I
have
this
nice
Demand
with
tags,
a
new
IDC
proxy
that
allow
us
to
do
authentification,
I,
don't
I,
don't
think
we
I
want
to
basically
close
all
the
options
basically
see.
What's
what
we
can
do
and
basically
decide
on
the
what's
the
best
best
outcome.
F
B
So
like
right
now,
only
a
few
folks
within
Sig
infra
actually
have
access
to
our
AWS
resources
and
that's
kind
of
fine,
but
it's
it's
not
necessarily
manageable,
as
we
want
to
start
bringing
more
contributors
on
to
be
able
to
manage
and
help
support
all
of
our
resources,
and
this
isn't
just
in
my
mind.
This
isn't
just
an
AWS
problem
like
this
is
something
that
whatever
we
choose
here.
We
want
to
also
be
able
to
use
as
we
expand
into
other
potential
cloud
providers,
so.
F
Okay
and
the
other
thing
here
is
so
further-
we
are
looking
for
a
POC
on
this
one
like
or
or
so,
basically
like,
not
totally,
all
your
these
tools,
based
on
the
I
mean,
come
on
the
outcome
of,
like
you
know,
to
have
this
pros
and
cons
of
this.
This
providers.
A
B
World
in
an
ideal
world,
what
would
happen
is
there
is
a
GitHub
team
in
the
kubernetes
org.
That's
like
Kate's
infra,
Release,
Me,
In,
kubernetes,
there's
release
manager.
It
would
be
like
Kate's
infra
managers
right
and
if
you
were
part
of
that
GitHub
team,
you
can
authenticate
using
GitHub
to
get
access
to
AWS
gcp
Azure
data.
C
Yeah
yeah
right
now
we
do
it
with
Google
Groups
because
it
works
well
with
gcp.
When
we
go
beyond
gcp,
we
need
options
for
the
other
clouds,
not.
A
We
want
to
expose
the
community
and
not
only
like
you
say,
not
only
AWS,
even
maybe,
if
we
get
Azure
or
if
you
get
Oracle,
we
need
to
be
able
to
basically
allow
and
future
in
the
community
to
access
to
spit
to
a
minimum
of
resources,
because
we
need
some
kind
of
transparency
at
some
level,
although
infrastructure.
So
the
idea
is
to
basically
build
a
uniform
platform
for
Authentication.
A
I
personally
for
SAS
solution,
because
that
reduce
the
burden
of
Maintenance.
So
if
you
can
get
induct
and
Doctor
license
or
license,
that's
a
plus
one
for
me
now,
if
you
feel
like
this
is
too
expensive
for
cncf,
we
look
at
now.
What
are
the
open
source
options?
Are
we
building
something
from
scratch
by
tradition,
or
are
we
using
existing
open
source
platform.
B
I
will,
at
the
very
least,
once
Chris
is
no
longer
jet
lagged
ping
him
and
see
if
he
has
any
contacts
over
an
OCTA,
because
Chris
has
connections
everywhere
other
than
that,
if
Muhammad's
interested
in
Azure
ad
or
messing
around
with
it
I
think
we
could
get
something
going
with
them
other
than
that.
This
might
be
something
I
just
kind
of
dig
into
and
give
my
two
cents
on
what
might
be
best
minus
the
whole
OCTA
and
Azure
thing
so
I'll
dig
into
it.
C
A
Yeah
we
have
an
issue
open
for
this,
so
we
just
I
think
they
were
like
different
conversations
around
this.
A
A
I
saw
Justin
working
on
basically
extend
that
tomorrow,
at
the
best
region.
I
think
that's
already
done.
G
There's
one
more
region,
so
we
didn't
do
all
the
regions.
We
could
do.
There's
one
more
region
that
is
the
next
region
is
Tokyo
or
something
that
is
like
high
impact.
But
I
was
sort
of
just
doing
the
regions
as
we
as
we
like
picking
them
off
the
top,
but
we
could
also
just
do
them
all.
But
I
don't
know
if
anyone
has
a
preference
yeah
what's
what's
missing,
is
we
don't
have
the
cloudfront
in
prod,
so
we
don't
have.
The
current
fullback
is
GCS.
Sorry.
G
Yeah
I
agree,
it's
probably
fine,
I
think
I.
Just
we
should
probably
do
something
we
just
haven't
it.
Isn't
that
we're
saying
we're
not
doing
it?
We
just
haven't
done
it
yet
and
I
think.
If
we
want
to
do
cloudfront,
we
have
to
make
sure
that
we
we
whatever
cloudfront
is
in
front
of
is,
is
always
populated,
so
I
need
to
move
the
populator
The
copier
into
the
poll
into
the
post
post,
submit
rather
than
a
periodic
in
in
prow,
so
that
is
always
populated
rather
than
only
every
hour.
C
C
I
know
one
sec:
can
we
go
back
to
Justin,
so
a
question
for
you
Justin,
the
the
artifacts
that
we
are,
the
porch
is
serving
right.
Are
we
planning
to
add
more
artifacts
or
or
when
are
we
gonna
like
right
now?
It
seems
to
be
for
cops
and
CRA
CRI
tools
and
a
few
few
chosen
things
right.
Do
we
want
to
do
more
and
what
would
it
look
like.
G
Well,
they
they,
we
didn't,
choose
them,
it's
whoever
opts
in.
So
it's
open
to
everyone,
the
the
big,
the
big
obvious
one
is
dl.kates.io
and
the
bucket
behind
that,
but
so
like,
for
example,
that's
where
people
download
cubelet
and
Cube
cuddle
from,
however,
that
one
is
not
on
this,
not
on
our
current
budget.
It's
sort
of
off
the
books
as
it
were,
because
it's
still
served
from
non-cncf
infrastructure,
okay,
okay,
so
if
we
were
to
move
it
over,
so
we
could
move
over
dl.kates.io.
G
There
are
still
people
going
direct
to
the
bucket,
so
we're
gonna
have
another
project
where
we
we
have
to
stop
people
going
direct
to
the
bucket.
So
if
you
have
us,
if
I
know,
chaos
has
it.
For
example,
if
you
have
a
storage.googleapis.com
URL
for
downloading
Coupe,
cuddle
or
cubelet,
that
is
not
not
good,
and
if
we
were
to
somehow
magically
move
the
bucket,
which
we
can't
do
anyway
like
that,
would
be
bad
for
the
cncf
budget
on
GCS.
G
We
can,
however,
move
dl.case.io
as
long
as
we
then
redirect
to
AWS
that
will
be
bringing
more
stuff
under
CNC
of
control,
but
that
is
probably
the
next
big
thing.
The
alternative
is,
of
course,
to
just
start
asking
the
release
people
to
start
publishing
to
to
artifacts.case.io,
but
it
really
doesn't
matter
whether
we
choose
to
stand
up
dl.movedl.kids.io
or
move
the
functionality
under
artifacts.
G
G
Think
images
are
the
top
priority
right
now
for
in
terms
of
the
reducing
cncf
Google
specs.
A
Okay,
so
this
is
more
like
request
for
review
like
gyms
draft,
a
proposal
to
replace
cluster
up
that
they
search
with
something
so
I
highly
suggest
for
people
to
put
all
the
thoughts
about
this
and
have,
if
you
have
any
question,
feel
free
to
reach
out
on
Slack
or
if
you
have
any
question,
we
can
address
them
now,
because
I
think
we
still
have
time
before
the
meeting.
C
And
this
is
something
that
we
have
to
run
by
everybody.
You
know
seek
testing
and
whoever
else
that
we
can.
Think
of
so
it's
not
just
us
making
the
decisions
here
but
I.
You
know
I
put
my
hat
on
as
a
long
time
person
looking
at
CA
jobs
and
it
it
came
from
there.
We
had
a
meeting
yesterday
with
some
of
the
qrimatic
folks,
looking
at
kkp
from
from
Cuba
Matic
as
one
possibility
of
hey
is
there?
Is
there
a
problem?
C
This
is
a
problem
that
we
we've
had
an
open
issue
from
2019
is:
can
we
even
solve
it
right,
then?
The
question
that
came
up
in
the
meeting
was
okay.
What
are
the
requirements?
Because
you
know
we
need
to
go
from
a
well,
you
know.
Well,
we
should
all
agree
to
it
and
then
we
should
look
for
what
we
should
be
able
to
do
to.
You
know
solve
the
requirement,
and
some
of
the
things
that
we
were
talking
about
was
hey.
C
We
have
chaops,
we
have
Cube
spray,
we
have
cap
Kappa,
we
have
a
lot
of
other
installers.
Why
haven't
we
still
not
moved
from
from
the?
You
know?
Bash
scripts
that
we
have
for
Cuba
and
a
lot
of
that
has
been
captured
in
the
in
the
docs.
So
please
leave
comments
there
hit
me
up.
We
can
do
it
on
slack
as
well.
I
posted
it
to
both
sick
testing
and
kids
infra.
So
we
can
use
the
threads
there
to
talk
about
things
as
well.
C
G
Yeah,
that's
that's
really
great
to
see.
Thank
you
for
sharing
I,
don't
know
if
you've
seen
I
put
a
link
to
something
I've
been
working
on
so
I
share
the
desire,
and
we
are
I've
started
with
cloud
provider
gcp,
because
we
found
a
bug
in
our
usage
of
cloud
provider,
gcp
and
Chaos,
and
what
we're
trying
to
do
there
is
basically
get
the
other
problem
is
I
think
we
want
each
project
to
be
able
to
do
these
tests.
So
I
will
comment
on
your
document
on
this
yeah.
C
C
A
C
Anything
one
more
thing
or
no
yeah
there
was
the
I,
don't
know.
If
we
should
talk
talk
about
it
here
or
in
the
other
meeting
the
fastly
one
Ben
had
mentioned
hey.
We
need
to
bump
up
what
we
need
from
fastly
in
60
petabytes,
or
something
like
that
right.
So
do
we
talk
about
that
here
or
do
we
talk
about
in
the
regular
meeting.
A
A
I
think
yeah,
the
the
only
update
is
like
we're
waiting.
We
need
CSF
to
sign
the
service
order,
okay
from
fastly
first
like
even
before
we,
we
basically
request
more
bandwidth
for
context.
Basically,
we
we
underestimate
the
band
with
me:
I
underestimated,
if
needed,
for
the
l.k.io
ivory
courses,
60
terabyte,
but
currently
the
current
bandwidth
is
more
than
five
petabytes.
So
I
talked
to
folks
from
fasting
eyes
for
10
petabytes.
A
They
basically
say
they're
gonna
try
to
get
the
upgrade
and
in
the
meantime
we
should
sign
the
service
order,
the
fastest
episode
of
send
to
cncf
logo.
So
I
think
that's
the
first
thing
we
need
to
do
like
sign
the
fast.
The
service
order,
finish
the
configuration
and
do
the
fonts
we
get
the
bandwidth
upgrade.
We
do
the
flip.
We
can
do
the
communication
and
do
the
flip
mm-hmm.
C
A
B
I,
recall
and
I
will
have
to
double
check,
because
this
is
where
things
got
things
stalled
a
little
bit.
Cncf
just
wanted
it
to
be.
B
Thank
you
to
fastly
Google
aw,
like
essentially
no
one
provider
like
donor
is
held
above
the
others,
whereas
the
original
contract
and
fastly
was
like.
You
have
to
have
a
vanity
fast
URL,
which
is
fine,
but
it
also
needs
to
have
like
their
logo
Giant,
and
there
were
very
specific
marketing
things
that,
just
in
general,
the
cncf
and
the
LF
are
like.
No,
so
thesley
was
fine,
with
changing
their
contract
that
you
know
it's
just
template
language
copy
paste
and
then
hash
it
out
later.
B
It's
just.
They
made
the
changes,
and
then
it
went
to
LF
contracts
to
do
their
whole
review
and
I
think
it
might
have
to
go
back
again,
because
if
the
contract
specifically
States
the
50
TB,
instead
of
the
50pb
fun,
no.
A
B
But
the
the
contract
still
States
the
like
base
donation
of
a
certain
amount
and
given
new
info.
We
now
need
to
make
sure
that
contract
has
the
order
of
magnitude
more
that
we
need
and
I
imagine
once
the
contract
has
finished
going
through
review.
That
change
won't
take
a
whole
freaking
cycle
of
contract
like
legal
review,
it'll
just
be
like
Yep.
This
number
changed
to
bigger
great,
but
yeah
I
I
will
poke
Joanna
and
see
where
that
is
because
that
is
100
in
her
court.
All
I
can
do
is
kind
of
continually
prod.
A
Okay,
there's
nothing
in
the
agenda.
We
want
to
talk
about
something
specific
and
I.
Give
you
14
minutes
of
your
time
back.
G
Oh
I
just
wanted
to
mention.
Thank
you
for
the
promise
Mario
we
were
talking
about.
I've
been
talking
with
the
kubernetic
folk
and
specifically
Patrick
about
looking
into
things
like
the
Amazon
controllers
for
kubernetes.
So,
in
other
words,
like
is
terraform
the
the
right
does.
Terraform
enable
us
to
have
a
collaborative
like
GitHub
based
model
for
deployment
or
is
would
something
else
be
better,
so
I
think
crematic
are,
and
Patrick
in
particular
are
checking
that
out.
C
G
A
A
A
Okay,
so
James
talk
about
basically
yeah
continuity
got
gcp
project
to
run
the
Italy
test,
so
I
would
like
to
basically
get
some
some,
basically
some
some
kind
of
access
to
the
gcp,
so
we
can
import
them
on
the
bus
gospel
run
by
kubernetes
and
flip
that
project
to
use
those
gcp
projects.
How
we
make
that
topic.
C
This
is,
let
me
give
you
give
the
context
to
everybody
else
on
the
chart.
Continuity
is
a
separate
project
continuity.
You
know
we
want
to
test
continuity
with
latest
tip
of
what
is
there
in
kubernetes
master,
and
so
there
is
some
shared
infrastructure
there.
Where
you
know
continuity
has
pre-summit
jobs
and
post
Summit
jobs
that
run
on
our
brow,
and
you
know,
for
the
longest
time
there
was
a
project
owned
by
Google
under
which
they
used
to
run
it.
Now.
C
The
continuity
folks
approached
cncf
and
they
got
a
new
gcp
pro
project
allocated
to
them,
and
we
want
to
tell
our
Pro
to
run
the
ca
jobs
that
belong
to
continuity
in
their.
You
know,
GCB
project.
So
that's
that's
the
background.
B
So
this
is
going
to
get
real
funny.
I
have
a
question:
how
long
ago
did
that
container,
D
or
get
created.
C
B
B
I'm
gonna
look
right
now,
so
there's
this
fun.
There's
this
fun
thing
right
now:
I
hippie,
ehore
Rob,
like
Chris
Amy.
We
all
have
the
highest
level
of
access
to
the
cncf
gcp
org,
and
none
of
us
can
create
a
new
project
that
gets
billed
against
and
we
have
had
a
ticket
open
with
Google
for
about
six
months,
hippie
longer
to
try
and
figure
out
who
the
real
owner
of
the
cncf
gcp
org
is
because
at
this
point
no
one
can
correct.
We
do
not
have
permissions
to
the
billing
account.
Okay,.
D
I
think
I
can
also
I
I
need
to
go
look
but
I
think
there
is
some
container-d
accounts
within
the
gcp
org
already
that
that
were
not
con
top
level.
Give
me
a
moment
to
look
that
up.
C
B
Rob
created
CRI
container
depr
node
e2e
under
the
cncf
yeah
gcp
org,
which
means
it
might
have
already
had
a
billing
account
and
we
would
be
good.
So
at
that
point
all
we
have
to
do
is
create
a
gke
cluster
and
pass
on
the
cubeconfig
for
y'all.