►
From YouTube: Kubernetes WG K8s Infra 2019-04-03
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
B
B
B
B
A
B
B
E
F
E
G
F
So
just
pulling
up
the
billing
report
for
the
last
30
days,
there's
not
specifically
the
last
two
weeks.
It's
the
last
30
days,
grouping
by
project.
What
we
see
is
DNS
has
cost
us
about
$40
38:29
in
particular,
and
the
compute
engine
has
cost
us
three
hundred
and
twenty-seven
dollars,
which
I
presume
again
is
Justin's
test
clusters,
clusters,
plural
and
hopefully
we'll
those
will
be
online
for
real
soon,
and
this
will
be
a
permanent
charge,
so
we're
using
them
for
real.
F
F
A
D
E
The
hot
stuff
has
been
merged,
the
proc
configuration
to
get
basically
the
the
untrusted,
like
main
Wow
instance,
to
look
at
kubernetes
flash
case.
I
oh
that
people
love
the
manifests
there
and
I've
already
started
testing
this
yesterday.
I'll
probably
finish
this
later
today,
assuming
there
are
no
additional
issues,
but
basically
it's.
E
E
The
main
remaining
items
are
like
a
testing
GCRs
for
setting
up
like
an
e3
test
framework.
Tim
you
Tim
Hawk,
and
you
mentioned
this
before
I
just
saw
your
plummets
on
a
PR
yesterday
and
replied
back,
but
basically
test
VCRs
for
testing
only
and
I
guess,
like
the
the
et
framework
itself.
That
would
be
really
nice.
I
already
have
an
issue
open
for
that.
I
can
link
it
in
the
dock
after,
but
I
would
think
that's
about
it.
I
mean
after
like
the
this
initial
like
demo
for
Kate's
IO
is
like
starts
working.
A
F
F
F
In
the
mean
time,
anybody
object
to
me
creating
a
staging
and
prod
just
for
promoter
testing,
no
GCRs
using
the
same
script
that
we
have
before
actually
I'll
have
to
mutate
this
script,
because
the
prod
script,
sort
of
assumes
a
single
name
and
so
I'll.
You
take
that
script
a
little
bit
to
create
the
promoter
prod
and
the
the
real
prod
and
we'll
need
a
google
group
that
governs
access
to
those
any
objections
to
that
all
right
Linus.
We
can
set
that
up
either
later
today
or
later
this
week,
cool.
F
E
One
more
update,
I
just
saw
in
my
email
heart
meat
club
from
VMware.
He
added
a
PR
for
the
promoter
to
snapshot
of
existing
GC
charge
so
that
we
can,
you
know
once
we
did
like
the
book
migration
thing.
You
can
just
take
a
snapshot
of
existing
old
prod
and
then
just
create
a
manifest
out
of
that.
You.
I
E
F
E
F
F
You
know
being
used
for
the
world,
isn't
in
any
state
rebo,
and
so
the
question
is:
do
we
make
a
phony
grandfathered
staging
repo,
where
we
upload
all
those
to
the
phony
grandfathered
staging
and
then
let
the
promoter
promote
from
there
or
do
we
try
to
create
staging
repos
that
are
legitimately
for
each
of
those
topics?
Oh
my
god,
or
do
we
just
load
them
into
the
new
prod?
E
F
F
Gonna
have
almost
no
time
between
now
and
our
next
meeting,
because
I
will
be
out
for
the
almost
the
entirety
of
next
week
and
then
the
week
after
that,
I'll
be
paying
off
that
so
I'm
reluctant
to
commit
to
anything
in
the
two-week
time
for
him,
but
I
will
keep
doing
my
best
to
push
this
forward
and
reviewing
whatever
PRS
Lana
sends
me.
Thank
you
see.
B
Okay,
excellent,
it's
good
to
see
the
image
before
we
decided
to
miss
everybody
going
to
the
DCP
thing.
Maybe
I
can
attend
some
time
after
our
image
promoter.
We
have
some
artifactory
a
bit.
B
D
B
The
occasion
have
so
we
have
a
CI
system
or
apex
nuke
that
CSPs
CI
I
I've
got
all
of
those
doing.
Http
URL,
but
I
wanted
to
set
up
something
for
on
page
sees
an
API
snooped
at
CN
CF
to
IO,
which
is
the
main
site
and
I
wanted
to
transparently
redirect
it
to
the
or
302
redirect
it
to
the
right
and
CIN
is
for
on
APA
suit.
That
gives
it,
let's
see
hi
so.
D
F
Do
if
it's
just
to
you
or
an
all
redirection,
you
can
send
it
as
a
PR
agency,
untaxed
kindig
in
the
case
that
io
subdirectory,
if
you
want
to
go
link,
sorry
I
was
looking
at
something
trying
to
hear
one
of
my
computer's
taking
off.
While
we
chat
is
the
sorry,
did
you
want
to
go
linker
a
top-level
okay
said:
I
have
a
URL.
F
If
we,
if
we're
confident
that,
were
that
we're
keeping
this-
and
this
is
a
real
thing
and
I
don't
object
to
that-
we
should
think
about
whether
we
want
the
name
to
reflect
the
implementation
or
the
intention.
But
that's
a
discuss
that
so
yeah.
You
can
just
send
a
PR
against
the
internet
and
think
that's
really
easy.
Awesome,
I'll.
B
Do
that
okay,
move
on
from
there
thanks
for
those
days,
Justin
I,
know
that
you
know,
probably
also
being
a
couple
of
weeks,
won't
expect
anything.
There
there's
a
lot
of
conversation
on
net
two
five
and
they
were
document
how
to
set
up
DNS
under
4k
kate's,
then
I,
don't
think
we
the
documentation
is,
is
there
and
we
were
running
someone
I,
don't
think
we
had
a
somebody.
B
Take
this
that
we
were
going
to
ball
until
sig
Knox
to
create
a
sub
project
for
an
l,
a--
find
of
hand
over
the
network
by
Porsche
new
sig
docks
and
make
sure
that
we
just
document
how
the
DNS,
within
the
case
that
I
was
there
an
offline
discussion
creation
issue,
template
creating
and
those
are
the
the
network
I
top.
Does
anybody
wants
to
speak
to
those
feel
free
to.
B
F
H
B
Look
for
a
pair
with
someone,
while
we
do
I'm
going
to
get
access
to
do
a
burn
down
and
create
cycle
of
a
test
cluster
so
that
we
have
more
people
with
that
skill
set
and
also
have
it
well
documented.
It's
something
I'm
passionate
about
is
pairing
with
people
who
have
great
skills
and
documenting
that.
So
in
the
process
of
pairing.
We
capture
that
for
the
rest
of
our
community,
so
I'm
balling
about
and
telling
people
to
invite
me
to
pair
with
them.
B
B
B
J
F
J
Specific
adverse.
The
question
where
we
start
is
that
who
is
too
sweet
because
she
contributed
experience,
has
G
sweet
account
but
its
regular
account.
It's
not
an
admin
account
and
some
questions
that
they
would
like
to
address
Google
suite.
But
if
you
have
the
regular
account
you
are
not
able
to
do
so.
F
J
F
B
A
B
E
E
A
PR
the
one
that
I'm
working
on
right
now,
yeah,
okay,
that
that
is
actually
the
demo
itself
folks
I
mean
I,
can
add
a
link
into
the
discussion,
but
I
mean
that's
like
the
real
thing.
That's
actually
you
can
see,
you
can
click
on
trial
results
see
the
you
know
like
the
shah's
and
digests
and
all
that
channels.
Alright,.
F
F
Miss
drinkin
here
Jonathan,
no,
he
is
so
I
can
talk
just
a
little
bit
about
it.
He
found
this
link
service,
which
seems
like
it
does
what
we
need
and
has
started,
putting
a
PR
together
to
do
it.
There
were
some
issues
with
the
PR
and
how
it
integrates
with
kubernetes
and
how
it
restarts
and
those
sorts
of
things
he
used
working
through.
That
was
the
last
I
heard
of
it.
A
couple
days
ago,
Friday.
H
I'm
so
I
did
the
suggestion
of
you.
The
DNS
face,
because
that
has
said
past
is
basically
review
and
then
because
of
the
new
DNS
infrastructure
and
I
will
be
in
town
infested
Cisco
for
next
next
week
and
I
have
a
meeting
to
discuss
the
full
request,
and
if
we
go
that
route
and
in
general,
what
we
can
do
on
Fridays.
So
then
we
can
see
how
that
goes
and
I'm.
Basically,
volunteering
to
help
add
on
the
redirect
aside
or
shorten
aside
and
depending
on
what
direction
we
go.
F
H
Pr
is
basically
using
the
I
think
the
slang
application,
which
is
using
a
database
in
the
backend
and
the
different
idea,
is
instead
of
having
a
specific
database.
We
reuse
the
DNS
zone
records
which
are
in
human
eyes
that
I
are
in
the
repository
and
basically
uses
them
as
a
key
value
store
and
go
based
application,
which
basically
is
a
stateless
thing
that
we
can
run
globally
depending
on
what
how
much
we
need
and
pull
in
the
specific
DNS
tax
records
or
the
values
from
the
key
value
store.
H
If
you
want
to
call
that
and
redirect
accordingly,
is
this
a
config
that
link
already
has
support
for
that's
a
different
thing,
so
one
is
tech
director
and
the
other
one
is
link
which
is
basically
using
the
database.
So
it's
two
different
things.
One
is
one
on
the
PR
and
the
other
one
is
just
suggestion
which
might
integrate
better
into
the
current
tooling
and
review
process.
H
F
Because
the
this
link
stuff
seemed
pretty
straightforward,
it
was
basically
edit
a
config
map
and
push
it
to
the
cluster
and
everybody's
happy
and
honestly,
it
probably
doesn't
even
need
to
be
back.
It
could
be
the
git,
repo
or
subdirectory
of
the
git
repo
that
we
could
watch
and
have
an
update,
but
just
link
itself
didn't
do
dynamic
updates
and
so
Jonathan
was
spending
a
bunch
of
time
banging
his
head
against
the
wall
of
how
to
make
it
recognize
file,
updates
and
restart
yeah.
H
So,
that's
probably
it
so
that's
with
having
basically
lunch
in
sunny
all
on
phrase.
So
we
wanted
to
see
what
his
approach
is.
What
the
other
option
is
not
basically
jumping
in
and
completely
saying
like
he's
finding
anything
yet,
but
mostly
you
have
a
discussion
on
what
works.
What
doesn't
work
with
the
currency
are,
and
if
we
want
to
reiterate
and
if
you
need
any
help,
I
can,
if
you
provide
any
way
awesome.
Thank
you.
That's
this
Friday
you're
gonna
be
here.
Yeah
I'm
flying
tomorrow,
I
think
yeah.
Okay,.
H
So
one
one
we
added
one
reason
to
basically
go
the
dns-based
route.
Is
the
delegation
nature?
So
currently
everything
would
go
through
one
central
repository
and
with
a
dns
based
approach.
We
can
basically
push
out
to
the
episode
which
can
be
managed
by
either
like
a
sick
or
anything
else,
and
that
could
be
basically
be
a
more
distributed
approach
instead
of
having
one
repository
and
one
maintainer.
Gene
basically
even
do
have
path,
and
something
like
that
which
might.
H
F
H
A
The
new
application
needs
well,
the
code
for
the
NGO
application
that
you
mentioned
will
lead
to
live
somewhere
right.
So
can
we
just
assume
that
it
lives
in
a
separate
repose
somewhere,
and
then
we
will
just
use
the
container
image
from
there
here
when
we
are
bootstrapping
stuff
here
we
don't
need
to
own.
The
code
also
know.
H
That
the
open
source
project-
and
that
would
be
similar
to
sling,
where
we
justify
our
ready,
docker
image
and
run
it,
and
the
only
thing
that
is
needed
is
basically
it
doesn't
need
anything.
It
might
have
made
a
Cordy
mess
as
a
caching
layer
or
something
like
that,
but
usually
you're,
not
really
it
that
even
and
it
does
out
of
TLS
and
stuff
like
that,
because
it's
based
on
Kenny.
So
it
has
a
lot
of
features
that
might
for
more
advanced
things.
If
that
might
be
the
need.
Okay,
yeah.
H
B
Changes
in
our
I
am
configuration.
The
PR
is
pretty
straightforward,
but
also
includes
a
dump
of
the
I.
Am
there's
a
little
bit
of
walking
documentation
on
how
to
initially
run
that
and
submit
a
PR
note
that
within
that
PR
there's
also
the
the
buckets
and
the
files
within
the
buckets
quick
reading
public
billing.
As
far
as
information
that
we
can
start
using
to
generate
detailed
billing
information,
we
go
over
an
org
policy
and
a
I
wasn't
sure
and
get
your
everybody's,
a
quick
feedback,
we're
going
gamble
or
JSON
dumping.
B
The
yam
will
shorter
and
but
the
JSON
half-dozen
one
of
the
other,
except
for
where
we
have
there's
no
roles
within
the
CN
CF
for
with
an
empty
list
and
it's
kind
of
confusing
to
have
a
llamo
file.
That's
that's
the
girl,
bytes,
so
I
it
was
leaning
toward
based
on
mainly
for
that
reason
for
the
kubernetes
public
or
project
we've
got
a
dump
there
and
it's
interesting
to
see
when
things
were
changed.
Well.
B
Temp
document
was
updating
things.
We
have
a
commit
that
includes
the
changes
he
made
during
that
pear
session
to
the
I
on
policy
and
I
suspect
that
as
we
move
forward,
whatever
whatever
is
polling,
there
will
help
do.
Polling
right
now
would
automatically
great
that
to
review,
and
maybe
some
way,
if
there's
a
way
that
we
can
audit
who
made
those
changes
and
assign
that
change.
B
Pr
to
the
person
who
made
the
changes
with
to
external
auditors
would
be
great,
some
some
other
additions,
I've
thought
of
where
instead
of
polling
with
a
cron
job
or
something
in
prowl
monitoring
those
changes.
So
it
triggers
a
review
of
the
policy
change.
So,
let's
just
upgrade
especial
and
I
appreciate
any
feedback
on
this
approach
and
where
we
can
go
from
here,
there
are
more
policy
things
to
want.
B
It
I
believe
I
put
them
at
the
end
of
the
audit
strip,
around
dumping,
the
im4
eats
PCs
bucket
and
the
ion
for
bigquery
and
iterating
over
the.
As
far
as
getting
every
possible
thing
that
could
be
built
querying
for
the
enabled
api's
for
the
kubernetes
public
project
or
any
projects
with
after
that
and
within
those
api
is
identifying
each
resource
that's
available
and
dumping
there.
I
am
thanks.
F
So
I
started
looking
at
the
PR.
It
looks
good
it's
unfortunate
that
it
has
to
be
reactionary,
that
it's
not
the
authoritative
source
for
for
configuration.
We
could,
in
the
future,
move
towards
it
being
the
authoritative
source
and
sinking
in
the
other
direction.
I
have
a
little
a
little
nervous
about
it
right
now,
because
there's
so
many
different
things
to
look
at,
like
you
said.
F
B
Well,
we've
got
a
lot
of
Googlers
on
here.
They
are
familiar
intimately
with
GCP.
Is
there
anything
in
auditing
where
we
could
have
something
said,
be
a
string
for
the
to
enable
are
almost
alert
based
when
there's
a
change
made
atomically
creating
that
PR
and
exciting
it
finding
who
made
that
change
so
that
that's
public
so
that
if
it's
a
bot
find
it
to
bond
if
it's
a
person
assigning
it
to
a
person,
so
in
conversations
can
be
transparent.
B
F
As
far
as
we
know
there
isn't
a
sort
of
watch
based
mechanism
for
this,
so
it'll
happen,
polling
based,
and
there
is
an
audit
log
I'm,
not
sure
what
the
API
for
accessing
the
audit
log
is.
I've
never
tried
to
do
it
through
the
API
I've
only
ever
done
it
through
the
UI,
but
it
is
there
something
we
can
look
at
and
figure
out.
If
we
can
link
to
audit
logs
back
to
I
am
Delta's.
B
I'll
have
to
think
about
that,
or
do
some
digging
into
it
to
see
if
there's
a
network
separately
from
the
watch
it.
It
might
be
interesting
to
know
who
who
made
those
changes.
So
in
that
audit
log
for
the
API
beam
indication
who
who
was
authorized
to
do
that
would
be
really
interesting
as
far
as
Auto,
creating
that
PR.
When
we
do.
F
That's
what
should
show
up
in
the
autobahns?
Oh
right,
the
audit
log
should
say
you
know
a
change
and
whenever
co-opted
I
know
whatever
email,
h/h
added
I
am
rule
whatever
that
and
that
should
be
in
the
audit
log.
The
only
question
I
really
have
is
kind
of
good.
You
should
be
able
to
see
the
auto
log.
Can
you
can
you
not
like
if
you
go
to
Monday
the
console
for
DUI,
can
you
not
speak
up.
F
F
F
F
D
F
A
F
F
B
A
K
A
B
L
K
Don't
know
that
so
I
guess
I
had
had
sort
of
two
two
paths
in
my
head:
one
that
maybe
we
would
improve
on
this
existing
google-specific
stuff
and
be
able
to
somehow
pull
people
in
to
have
more
hands
on
deck
versus
build
something
that's
open
and
transparent,
community-driven
on
the
new
infrastructure
and
I.
Think
at
this
point,
I'm
not
trying
to
like
tilt
it
the
window
of
opening
up
this
google-specific
thing
versus
just
let
it
be
how
it
is
and
for
114
we
did
better.
We
understand
some
of
the
lags
and
manage
that
better.
K
K
Yeah
I
I
think
we
have
I,
wouldn't
call
it
a
fire,
but
we've
got
a
simmer
and
we're
talking
about
trying
to
get
the
other
stuff
rolling
through
the
115
cycles.
So
I
guess
from
my
perspective,
it
depends
on
all
these
other
things
that
are
being
stood
up.
The
the
packaging
feels
like
a
relatively
minor
one
to
move
at
the
point
that
these
other
things
are
working
out
and
I
don't
want
to
complicate
the
basic
agenda,
there's
like
a
huge
span
of
different
service
topics
right.
K
K
You
yep
and
in
there
I
think,
there's
also
one
of
the
things
that
slowed
things
down
a
bit.
We
had
a
realization
that
we,
the
way
we're
running
the
repos
today,
actually
causes
some
well
running
the
repos
and
managing
the
the
package
specification
content,
their
dependency
requirements
strings.
We
actually
have
broken
some
use
cases
and
what
the
right
thing
to
do
is
we're
still
hashing
back
and
forth
right,
so
there's
there's
hosting
content
and
then
there's
what
is
good
content
and
we've
really
circled
back
around
to
wait
a
second.
What
is
good
concept
host
sounds.