►
From YouTube: Kubernetes - kops office hours - 20190315
Description
Kubernetes kops Office Hours
kops: http://github.com/kubernetes/kops
A
Hello:
everyone:
it
is
Friday
March
15th.
This
is
cop's
office
hours.
I
am
your
moderator,
just
in
Santa
Barbara
I
work
at
Google.
This
meeting
is
being
recorded
and
will
be
put
on
the
internet.
So
please
respect
our
code
of
conduct
and
remember
or
don't
say
anything
you
wouldn't
want
put
on
in
public.
A
A
Two
images
so
far
are
about
how
we
do
our
Docs
and
about
the
burndown
and
I
thought
we
could
do
a
we
could
like
I
can
show
people
how
we
do
that
the
triage
would
have
to
socialize
some
of
the
hard
decisions
assume
the
responsibility
for
them,
not
just
not
just
actually
doing
them
and
then
but
gonna
give
him
visibility
into
that.
So
the
first
step
Peter.
Do
you
want
to
talk
about
Docs.
B
A
A
That
generation
is
basically
on
maintained
or
the
tools
to
generate
it
or
basically
are
maintained,
and
someone
created
a
good
tool
to
or
ala
DBF
I
think
submitted,
something
which
is
how
nginx
ingress
or
ingress
nginx
is
doing.
Their
docks,
which
is
it
I,
think,
is
more
similar
to
kind
and
I
agree.
We
should
do
something
better.
There
are
two
aspects:
I
guess,
there's
I
guess
are
three
aspects:
there's
generating
the
markdown
itself,
which
is
and
producing
HTML
there's
uploading
the
HTML
somewhere,
which
could
be
github
pages
or
could
be
I.
A
Think
kind
is
using
some
service
whose
name
I
forget
but
a
another
service.
We
should
probably
do
that
because
that's
whatever
else
is
doing,
and
the
third
element
is
figuring
out
what
to
do
about
our
generated.
Api,
Doc's
I
think
we
get
a
long
way
just
by
doing
the
first
or
starting
with
the
first
two.
A
Yes,
it's
it
would
be
great.
I
thought
I
saw
what
was
somewhere
to
do
it
without
so
the
issue
right
now
is
in
order
to
generate
the
Doc's.
We
have
to
generate
an
open,
API,
spec
and
then
it's
just
complicated
and
I
thought
I
saw
a
PR
similar
to
bypass
that
complexity
and
I
will
try
to
dig
that
up
but
yeah.
If
anyone
wants
to
look
into
how
kind
is
doing
it
or
something
like
that,
that
would
be
wonderful,
but
it
is.
It
is
a
problem
we
should
address
it
yeah.
A
C
I,
don't
remember
what
it's
called,
but
I
know
kubernetes
kubernetes
uses
it
and
that's
how
they
have
each
version
of
you
know
previous
versions
of
the
kubernetes
api
is,
you
know
documented
as
well,
but
I
I
think
this
is
like
I
agree
when
Peter
brought
this
up,
I
was
like
oh
yeah.
We
got
a
this.
Is
this
is
kind
of
like
a
a
thing
that
just
always
it's
kind
of
been
there
awhile
and
yeah
I
think
it'd
be
good
for
us
to
dive
into
taking
a
look
at
I.
C
Don't
I
hadn't
remembered
what
men
well
I
think
his
name
is
had
contributed,
but
now
that
you
post
that
Dustin
I,
remember
so
yeah
I
think
we
should
I
think
we
should
push
on
that
and
I
mean
I'd,
be
happy
and
maybe
Peter.
Maybe
you
and
I
like
work
towards
getting
this
a
little
bit
better.
I,
don't
know,
there's
a
lot
of
other
people
on
the
call
that
are
a
little
more
deep
divers
Justin
on
everything.
A
Be
great,
thank
you,
Peter
perfect,
and
definitely
we
can
take
this
help
want
it.
This
is
one
that,
like
people
may
get
may
contribute
to,
but
that'd
be
great
and
I
would
try
to
get
to
it,
but
I
am
focusing
on
112
or
trying
to
focus
on
112
right
now,
as
the
priority
I
think
that
should
be
a
bronze
priority
for
yes,
okay,
good.
A
If
there's
nothing
else
on
that,
I
was
going
to
do
a
burn
down,
so
we're
burning
down
for
112
there's
about
40
ish
PRS
that
are
open
that
are
candidates
for
112.
Not
all
of
those
has
to
go
in
before
112
out
for
one,
but
some
of
them
should,
and
some
of
them
will
get
punted
and
I
thought
we
could
do
a
like
triage
event
together.
If
there's
nothing
but
I
also
wanted
I
want
to
check
that.
That
sounds
valuable
to
people
and
also,
if
there's
anything
else,
we
wanted
to
talk
about.
First.
A
A
A
A
Is
this
working?
Can
you
see
a
cups
github
window
with
weekend?
46
open
issues?
Okay,
so
this
is
basically
how
I
do
the
triage
of
issues,
sort
of
PRS
Xi's
are
separate,
so
I
do
a
filter
is
PR
is
open,
is
pretty
much
normal
and
then
I
disregard
ones
that
are
totally
not
ready
or
not
not
ready.
Like
there's
nothing.
We
can
do
about
them
so
like
if
they
don't
have
a
CNC,
F
label
or
CN
CF
CLA
approval.
Then
you
know
we
ping
that
issue,
but
it
can't
go
into
a
it's,
not
something.
A
We
can
really
consider
to
merge.
I'm
sorting
by
updated
ascending,
so
I'm,
going
from
the
oldest
to
the
newest
and
I
also
exclude
I
started.
Creating
this
label
called
waiting
for
input,
which
means
that
sort
of
the
ball
is
in
someone
else's
court
like
they're,
the
PR
isn't
ready
to
merge
and
we
need
the
PR
submitter
to
do
something
and
that
can
include
me
I
think
I
have
I,
have
assigned
myself
waiting
for
input
on
occasion.
A
A
Should
have
nothing
in
in
a
milestone
other
than
112
or
113,
so
nothing
no
milestone.
Alright,
that's
a
lie.
Okay,
hopefully
these
are
pretty
new
or
sins.
Yes,
okay,
anyway,
I
try
try
to
check
that,
like
things
are
in
the
right
milestones,
so
backlog,
the
milestone
is
typically
things
that
are
waiting
for
or
things
that
are
just
like
when
we
get
to
them.
Not
really
milestone.
Associated
Doc's,
often
go
into
backlog
if
I'm
not
ready.
A
113
is
things
that
we're
not
ready
to
do
in
112
or
are
not
important
like
that
type
of
fix,
but
I
think
like
Mike,
your
switch
to
go
111
one.
We
can't
actually
do
until
kubernetes
goes
to
111
one.
So,
for
example,
that
goes
there
and
then
things
that
are
like
longer-term,
like
increasing
the
bundles
or
htps,
which
is
a
sort
of
mirroring
type
issue.
A
A
Oh
yes,
so
I
actually
excluded
everything
that
was
not
in
the
113,
so
I
actually
have
backlog
in
this
list,
but
that's
okay
and
then
I
typically
go
through
and
like
look
at
these
and
try
to
understand
like
do,
they
have
to
go
into
the
112
release
and
in
particular,
do
they
have
to
go
into
112
alpha
one,
or
can
they
be
added
later
so
like
Docs,
for
example,
because
of
our
because
of
our
own
version?
Doc's
are
not
I
did.
A
A
So
that's:
let's
look
at
one
of
these
so,
for
example,
the
good
ones
I
here
I've
actually
got
me
tea
so
far,
so,
for
example,
this
one
move,
ntp
and
miscellaneous
passing
packaging
stations
code
is
one
of
mine.
That
is,
that
is
just
a
cleanup
that
one
actually
is
not
a
blocker
in
any
way.
So
I
might
just
move
that
one
immediately
to
the
next
milestone.
A
And
then
we
should
be
down
to
45
of
them:
yeah
a
45,
okay,
one
one
closer
and
so
like.
Let's
look
at
a
less
less
trivial
one,
so
like
this,
one
I
think
we're
waiting
on
on
Mike
I.
Think
but
that's
that's
not
I,
don't
think!
That's
a
blocker
either,
let's
get
to
some
trickier
ones,
which
is
really
where
we
want
to
be
so
flag.
Mappings
are
generally
safe
and
we
try
to
do
those
as
soon
as
possible.
But
this
one
looks
like
it
needs
a
rebase.
A
The
the
ones
are
super
luxe.
Next
are
more
interesting.
I
think
these
are
the
ones
that
I
think
are
them.
The
things
we
have
to
get
in
so
calico
is
switching
to
CR
DS,
an
ED
CD
manager,
that's
sort
of
completing
the
big
thing
for
a
roadmap
for
getting
to
Etsy
d3,
so
I.
We
have
to
do
that
by
113
and
so
I've.
That's
why
I've
marked
them
as
blocks
next
for
112.
A
They
are
sort
of
the
critical
features
that
we
have
to
get
in
and
then
we
have
to
fixes
that
are
basically
like
cleanup
fixes
for
other
PRS
that
got
in
so
like
we
revamped
how
docker
package,
names
and
versions
worked
and
I
think
some
of
the
tests
found
that
some
of
the
names
were
like
typo'd
or
some
of
that.
So
like
things
like
that,
where
it
was
important
to
get
the
feature
in,
but
we
needed
something
else
as
well
same
thing
with
here.
A
This
is
the
ATSC
NIV
piece,
eight
of
us
V,
PCC
and
I
plugged
in,
and
we
wanted
to
supply
it.
Everything
on
do
a
to
a
man,
a
bundle
bump
as
I
recall.
So
just
that
sort
of
thing
and
then
the
final
one
is
another
one,
similar
sort
of
cleanup,
one
where
we
have
an
API
change
that
went
in
in
this
in
between
111
1
and
112
0
or
hasn't
yet
been
released.
A
And
maybe
we
want
to
like
tidy
up
the
API
a
little
bit
because
some
of
the
data
types
and
so
that
one
we
basically
have
to
merge
it
now
or
not
at
all,
because
once
it's
released
it's
done
and
it
looks
like
that
is
probably
blocked
on
me.
Getting
the
fixing
it
up
to
pass
tests.
Let's
look
at
some
other
ones
enough.
Anyone
has
any
questions
at
this
point
or
wants
to
lobby
for
anything
in
particular
yeah.
A
So
this
is
what
I
am
doing
and
I'm
trying
to
get
everything
into
112
alpha,
one
that
doesn't
disrupt
the
release,
elicit
let's
try
to
find
something
else
that
could
potentially
disrupt
the
release.
So,
for
example,
this
one
of
mine
which
moves
moves
and
configuration
files
around
it's
not
critical.
It's
helpful
for
112
a
TOEFL
for
GCE,
it's
not
a
blocker
for
112
and
I
could
probably
pump
that
one.
A
Similarly,
this
building
run
see
patch
I
think
that
was
a
alternate,
an
alternate
workaround
for
the
CVE
and
I
think
we
can.
We
can
we
can
pump
those,
but
that
one
as
well,
because
we
are
gonna-
do
oh
yeah
I,
remember
I,
didn't
pens
it
we're
gonna.
Do
we're
just
going
to
use
the
this
change
attribute
thing
instead
place
I'm
gonna
leave
I'm
gonna.
Leave
that
one
in
there
actually
for
now.
A
A
There
we
go
so
here
we
had
some
of
the
really
hard
calls.
This
one
has
been
Evelyn
for
way
too
long
and
it
changes
the
way
that
or
changes
the
DNS
record
that
gets
deleted.
So
it's
pretty
high
risk,
but
it
is
a
bug
fix
and
at
some
point
we
need
to
get
it
in
and
then
that's
where
I
sort
of
start
to
wonder
whether
we
should
put
it
in
now
or
put
it
in
later.
A
A
Probably
run
all
right,
it's
more
complicated,
I
thought
there.
This
is
area,
so
it's
around
the
weather.
We
assume
the
name
is
API
or
whether
we
assume
the
name
is
whatever
you
specify
on
your
master
public
name
and
like
whether
people
are
using
that
with
a
name.
They
want
a
keyboard
or
not.
It's
only
deleting
the
records
in
any
case
and
we're
already
obliterating
the
records.
So
this
is
probably
safe.
A
A
Sometimes
the
PRS
will
stay
open
for
a
long
time.
I
think
we
have
one
that's
a
year
old
and
I
do
feel
bad
about
that,
and
but
that's
sort
of
why
I
try
to
make
sure
we
look
at
all
of
them
and
that's
why
it's
not
a
trivial
exercise
to
do
this.
They
don't
new
release,
but
we
could
also.
We
could
adopt
a
different
approach
where
we
we
say
if
your
P
I
didn't
make
it,
it
didn't,
make
it
and
we
just
sort
of
release
when
it's
done,
which
is
more.
D
A
A
The
way
I
think
about
this
one
is
that
the
consequence
of
the
bug
is
pretty
low.
Like
we
Lika,
we
leak
a
DNS
record
where
you're
specifying
it
manually,
which
is
not,
in
other
words
we
if
you're,
using
if
you're
overwriting
the
public
name,
we
will
wash
I,
guess,
there's
two
pieces
so
we'll
leak,
the
master
public
name,
which
is
not
a
particularly
high
consequence.
I,
don't
even
know
if
there's
a
charge
for
records.
A
If
there
is
it's
very,
very
low
and
no
and
there's
a
limit,
but
it
seems
unlikely
that
this
is
gonna
push
you
up
to
the
limit.
I
guess
the
the
real
gotcha
would
be.
If
your.
If
your
API
name
was
a,
was
something
you
did
care
about
and
we
went
and
deleted
it
and
I
don't
know,
that's
actually
what
happened
here,
whether
it
whether
it
was
like
should
clean
up
whether
it's
please
don't
delete.
A
C
Especially
for
a
release
or
two,
you
know
we
could
also
try
to
you
know.
Once
we
get
up,
you
know
caught
up
a
little
bit,
then
maybe
we
slow
things
back
down
for
one
release.
Just
to
you
know,
get
a
bulk
of
stuff
in,
but
I
think
we
have
enough
demand
right
now
for
112,
okay,
I
think
it's
fine
to
bypass
a
few
things
personally,
I
figured.
D
A
A
C
A
C
C
A
If
not,
if
even
if
you
type
cone
bug
recurring
feature,
it's
still
useful
things
like
and
then
now,
even
if
I
have
to
apply
the
label
manual
is
something
to
the
world
alright.
Well,
that's.
That's
really
good
I
think
I
think
that's
that
will
help
get
112
out
faster,
then
and
I
hope
that
gives
some
clarity
I'm
like
when
I
think
things
should
block
next
and
why
and
sort
of
what
I'm
doing
here
and
yes,
if
people
want
to
help
that
is
wonderful.
A
C
E
Yes,
so
my
name
is
some
me
night
and
I
wanted
to
ask
a
couple
of
questions:
I
have
a
PR
for
M
GCP
private
network
topology
improvements
except
theirs.
I
am
really
not
familiar
with
the
development
process
and
it
doesn't
seem
to
be
much
of
documentation
around
it.
Let
me
find
the
link
to
the
PR
and
share
it
with
them.
Would
you
and
I
have
it
here?
You
just
open
it
a
few
hours
ago.
Yes,
I
did
all
right
there.
It
is
now
yeah
I'd
be
happy.
E
A
A
E
E
A
Months
going
twice,
alright
well,
I
wish
everyone
a
happy
weekend,
happy
Friday
I'm
happy
weekend,
and
we
will
see
you
all
in
two
weeks
and
hopefully
we
will
have
112
long
before
then,
but
yes,
yeah.
Thank
you.
Everyone
for
the
hugs
versus
features,
I,
guess,
decision
of
on
triaging
I
think
that'll
make
life
a
lot
easier.
Yeah.