►
From YouTube: Kubernetes SIG Windows 20190115
Description
Kubernetes SIG Windows 20190115
A
A
B
A
A
A
Yesterday,
a
lot
of
the
text
is
similar
to
to
what
I
had
started,
but,
of
course,
it's
been
updated
to
match
the
implementation
that
was
that's
been
discussed,
discussed
with
sig
off,
so
I
have
links
to
both
of
those
there
in
the
meeting
notes,
but
over
the
next
couple
of
weeks
our
goal
needs
to
be
to
get
the
rest
of
the
details.
They
are
needed
in
those
finalized.
A
Ll
GTM
a
PR
that
takes
it
from
the
provisional
to
the
implementable
state,
which
is
basically
saying
the
design
is
closed.
We
all
agree
on
what
the
what
the
contents
of
that
are,
which
includes
testing
and
release
criteria,
so
any
any
any
questions
on
that
in
that
process.
Okay,
now
that
you
find
your
room
Aaron,
do
you
want
to
give
some
more
details
around
the
release
process?
Sure.
B
Okay,
yeah
I
used
Brady
Bunch
boat,
so
I
can
see
what
you
all
do,
plus
it's
more
fun.
That
way
you
can
see
the
facial
reactions
so
yeah
hi
I'm
here
in
lieu
of
hi
I'm
Aaron
of
cig
beard.
I,
am
your
114
release
lead
I'm
here
in
lieu
of
Claire
Lawrence,
who
is
your
114
enhancements
lead
she
was
double
booked
today
we
are
at
week
two
in
the
release
cycle
week.
Two
is
traditionally
when
we
start
going
out
to
SIG's
and
asking
hey.
What
are
you
planning
on
working
on?
B
This
is
so
we
can
kind
of
have
that
plan
represented
as
enhancements
issues
that
are
in
the
114
milestone,
and
then
we
can
kind
of
lock
down
what
that
plan
is
at
enhancements,
freeze,
which
is
the
end
of
the
month.
Tuesday
January,
29th
and
yeah
Patrick,
I,
think
kind
of
has
been
keeping
his
eye
on
the
ball.
Since
a
lot
of
this
I
think
started
from
lessons
we
learned
during
113.
B
You
can
copy
and
paste,
sadly,
no
I,
don't
this
is
kind
of
going
to
be
an
iterative
process,
as
we
figured
this
out
as
a
community,
but
I
feel
like
I
just
saw
enough
vigorous
of
nodding
of
heads
when
I
have
said
this
statement
over
the
past
couple
of
months.
That
people
agree
that
it's
it's
time
that
we
do
this
so
I
am
planning
on
discussing,
let's
say
p.m.
later
today
to
look
into
like
adjusting
the
template
and
we've
like
we
might
have
to
adjust
it
a
couple
more
times,
but
I
do
think.
B
B
Traditionally,
this
is
also
when
the
enhancements
lead
would
ask
you
about
stuff
that
it
was
in
the
previous
milestone.
You
don't
have
any
enhancements
issues
in
the
previous
milestone,
so
I
don't
have
to
do
that.
I
would
ask
you
about
enhancements
issues.
You
have
the
V
114
milestone,
but
I
think
Patrick
just
did
a
good
job
of
numerating.
A
C
B
A
A
A
Yeah
I'll
take
a
closer
look
at
that
one,
but
my
initial
thought
is
that
it's
overly
broad
and
we
should
probably
cover
the
ones
that
are
going
to
work
and
and
one
of
the
ones
it's
called
out
on
their
security
contexts,
and
that
is
the
only
field
in
security
context
is
the
one
that's
covered
in
the
GMs,
a
cap
that
we've
already
got
linked
there.
So
I'll
take
another
look
through
it
there
and
it
may
be
that
we've
sort
of
piecemeal
this
one
apart
and
don't
need
the
master
issue
anymore.
I
think.
C
That
is
really
broad
and
I'm.
Now
we're
over
knee
in
the
urgency
to
go
through
the
config
and
just
isolate
Windows
ones
and
Donegan
said,
but
I
don't
think
we
have
been
with
to
tackle
that
disclosure,
so
maybe
we'll
deal
with
them
case-by-case,
for
example,
like
the
Jim
as
I
just
mentioned.
Now
we
can
ask
something
if
it's
needed,
but
maybe
not
go
through
the
entire
config
and
try
to
separate
windows
out.
D
B
D
B
Team
can
just
make
the
call,
rather
than
us,
having
to
adjust
and
iterate
on
the
goal
posts
as
we
go.
So
part
of
me
feels
like
maybe
it's
unrealistic
to
have
those
goal
posts
set
in
stone
by
enhancement,
trees,
but
that
we're
gonna
probably
aspire
to
have
them
set
in
something
resembling
stone
nearer
to
code
freeze
so
that
we
can
collectively
agree.
Yes,
this
is
worth
it
or
no,
this
isn't
worth
it.
I,
don't
know
that.
A
I
would
like
to
ask
for
clarification
on
upgrade
downgrade
plan.
My
understanding
in
the
previous
discussions
was
that
was
mostly
focused
around
when
there
are
things
that
you
know
changed,
add
or
remove
an
API.
Do
we
have
a
plan
for
being
able
to
upgrade
components
in
the
right
manner
and
I
haven't
seen
anything
saying
so
I
mean
like
sig
windows,
you
know
was
basically
formed
effectively
out
of
sig
note
in
order
to
create
Windows
node
support.
So
when
it
comes
to
upgrading
and
down
creating
the
cluster,
I,
don't
see
there
being
any
cluster
level.
A
B
So
again,
that's
a
really
great
question
that
I
don't
have
a
concise
answer
for
the
reason
I'm
asking
about
upgrade
count
downgrade.
This
is
another
process
scarring
or
lessons
learned
from
1:13,
where,
when
CSI
was
going
from
data
to
stable,
it
was
uncovered
that
switching
from
one
version
of
the
drivers
to
the
other
would
result
in
a
complete
loss
of
access
to
volumes
that
you.
So
if
you
were
on
the
ga.
D
B
Of
it,
you
would
lose
access
to
the
beta
version
of
it,
so
I
I'm
not
sure
if
I'm
looking
for
like
a
test
that
automatically
does
upgrade
or
downgrade
but
like
what
are
the
considerations.
What
are
the
risks?
What
are
the
things
you
can
and
cannot
support?
You
might
think,
like
I,
what
what
is
in-place
upgrade
even
mean
for
Windows
nodes.
This
is
stuff
like
that
great.
A
A
B
Ok
cool
so,
like
I
said
hi,
I
am
your
police
lead.
You
can
find
me
a
spife
XP
on
github
on
slack
at
gmail
comm
at
google.
Calm
is
the
only
one
that
notifies
me
during
business
hours,
but
I
am
here
in
lieu
of
Claire
Lawrence,
who
is
the
inheritance
MIT's
lead
for
this
release,
so
you'll
be
hearing
more
from
her
when
it
comes
to
shepherding
these
things
around
I
really
appreciate
you
all
taking
the
time
to
talk
with
me
today,
thanks
so
much.
A
So
list
here,
I
had
a
project
management
topic
that
was
talking
about
the
we've
basically
been
through
multiple
iterations
of
how
we
are
tracking
stuff,
starting
from
initially
Michael
had
a
road
map,
doc
that
was
started
and
I,
don't
know
2016
or
2017
that
sort
of
listed
all
the
different
features
and
considerations
that
were
there
for
Windows
of
what
once
I
joined
I
wanted
to
start
trying
to
split
some
of
that
stuff
up.
So
we
could
understand
what
was
effectively
like
in
the
works
and
what
had
pr's
going
in
github
and
so
github
projects.
A
You
know
what's
in
progress,
what's
still
on
the
backlog
and
what's
completed,
so
that
would
give
people
more
of
a
dashboard
type
view,
and
so
we
would
still
need
to
track
everything
or
so
we'd
still
need
to
tag
everything
as
14
milestone,
so
it
works
with
the
other
release.
Queries.
The
question
for
for
I,
guess
for
within
our
sake,
is,
would
having
a
third
of
14
board,
be
helpful,
or
should
we
just
simply
use
the
queries
in
github
and
look
at
the
count
of
things
until
it's
zero?
A
A
E
F
A
A
F
A
So
I'll
leave
this
on
the
agenda
for
next
week
because
it
sounds
like
like
so
I
copied
stuff
from
the
board.
I
did
know
prioritization,
so
let's
sync
with
Michael
who's,
the
other
chair
and
and
then
I
guess,
let's,
let's
give
let's
give
prioritization
to
try
on
that
stuff
next
week,
maybe
aim
to
get
the
top.
You
know
10
or
12
most
important
things
there,
although
I
think
that
if
we
were
to
pick
the
top
10
or
12,
hopefully
there
are
things
that
are
already
planned
for
we're
tagged
for
14,
but
we
could.
A
D
A
G
G
The
approach
we
used
was
using
net
Sh
in
Windows
and
just
forwarding
like
creating
a
tunnel
to
the
pod
IP
and
that
won't
work,
since
you
know,
if
it's
bound,
if
the
process
is
bound
to
localhost,
it's
not
listening
on,
though,
like
the
pod
IP
interface.
So
it's
just
curious.
If
anyone
here,
like
I,
reached
out
and
slack
and
didn't
get
any
replies,
I
was
curious.
G
A
G
G
G
A
Sort
of
to
clarify
here
so
tube
control
port
forward
is
often
used
by
devs,
and
so
I
can
be
on
a
remote
machine
where
I've
got
a
cube,
config
and
so
let's
say
I've
got
a
container
running
over
my
kubernetes
cluster
and
it's
got.
You
know
the
node
or
the
dotnet
debugger
running
within
that
container.
It's
bound
to
a
port
owned
by
that
container.
But
what
I
want
to
do
is
connect
from
my
local
machine.
A
A
G
Can
I
don't
know
like
make
some
kind
of
example
and
share
it?
Offline
like
in
slides,
can.
A
Yeah
my
initial
thought
on
this
is
that
we
could
probably
just
add
some
goal:
encode
in
the
cube
lid
itself.
That
could
find
the
pod
IP
on
that
local
machine
and
the
port
and
just
bind
directly
to
it.
So
we
could
basically
adjust
the
firewall
rules
where
it
doesn't
allow
external
connectivity,
but
it
does
allow
the
cubelet
to
talk
directly
to
the
containers,
IP
and
then
just
forward
that
packet
force
as
override.
H
H
H
A
H
D
H
A
A
A
Okay,
if
you
look
at
the
meeting
notes
from
last
week,
we
have
a
link
to
Sasquatch
repo
I,
just
pasted
it
in
there,
but
he's
working
on
steps
there.
The
current
problem
is
that
the
networking
doesn't
work.
So
if
you
follow
the
steps
he
has
they'll
get
you
as
far
as
being
able
to
start
a
container.
Just
the
networking
is
broken
for
now
and
so
I
think
the
ETA
was
around
the
end
of
this
month
and.
A
H
E
H
H
Yeah,
so
we
are
we,
we
are
doing
this
change
for
container
D
as
of
now
right
and
there's
nothing
stopping
for
us
to
do
the
docker
ship
change
as
well,
but
we
can
do
it,
but
we
can't
remove
that
code.
We
have
to
support
previous
versions,
so
we
will
keep
that
that
code
for
previous
versions,
but
our
spy
on
top
of
it
begins,
remove
it
I
think
just.
H
A
That's
correct,
M,
and
so
we've
been
discussed
over
the
last
couple
meetings
is
that
we
were
gonna
focus
on
2019
only,
but
before
we
can
go,
remove
those
workarounds
for
the
previous
versions,
based
on
what
we
just
heard
there.
We
would
need
to
still
make
some
changes
to
docker,
shame
and
then
get
the
C&I
plugins
updated
so
that
we
don't
break
our
existing
tests.
So.
C
A
H
E
A
All
right
all
right!
Well,
thanks
everyone
and
have
a
great
day.