►
From YouTube: Kubernetes SIG Windows 20180320
Description
Kubernetes SIG Windows 20180320
A
A
The
folks
who
give
us
on
own
work,
given
us
Patrick,
has
also
asked
to
a
little
bit
about
the
identity
under
the
containers
and
some
proposed
changes
to
ensure
that
group
managed
accounts
are
supported
for
basically
running
more
sophisticated
workloads
in
the
industry,
I'm
a
huge
supporter
of
that
change
and
then
I
believe
we
don't
have
any
Piazza
outstanding.
At
least
I
haven't
seen
any
updates
lately.
A
So
if
there
are
any
I'd
love
to
hear
an
update
on
that
and
then
if
we
can
also
get
an
update
on
the
testing
effort
and
our
meeting
with
seek
testing
at
1
p.m.
Pacific
today.
So
if
there's
any
updates
ahead
of
the
meeting
by
by
the
folks
at
cloud
base,
that'd
be
great.
So
so
let's
get
started
Bob
any
updates
from
you.
B
A
D
D
A
D
I
informal
thing
because,
basically
just
tell
them
hey
we
doing
this
to
cube
test.
This
is
what
we
saw
in
the
images.
This
is
how
we
plan
on
supporting
the
test
where
they,
where
it
is
me
that
this
is
the
issues
we
encountered
and
just
have
a
back-and-forth
SEC.
What
they
say
see
what
they
think,
because
most
of
this
effort,
seeing
by
the
looks
of
the
peers,
I
already
submitted,
is
basically
go.
She
asian
with
them
and
see
what's
the
best
approach
for
it,
the
one
one,
they
will
accept
someone
long,
a
quick.
A
D
E
So
the
idea
is
that
there
would
be
a
subscription,
I,
think
I
think
on
now
and
the
call
as
well
yes,
here,
okay
I,
prefer
that,
whether
to
say
that
we
have
address
a
subscription
that
would
be
then
be
used
by
Prowler
to
to
own
the
test.
I,
don't
know
if
Patrick
wants
to
answer
more
on.
A
This
I
think
one
of
the
things
that
a
cig
testing
is
going
to
ask
is:
to
what
extent
would
the
bigger
kubernetes
testing
infrastructure
we'll
be
able
to
use
the
same
subscription
for
those
tests?
I
think
that
would
be
something
that
we
may
have
to
have
an
answer,
because
they
will
definitely
have
an
asked
us
that.
B
Okay,
yeah
yeah,
yeah
I
can
follow
up
on
that
and
see.
You
know
what
the
desired
scope
is.
There
I
think
one
of
the
things
we
still
need
to
decide
is
whether
or
not
we're
going
to
test
every
single
PR
or
do
something
more
akin
to
like
a
daily
but
set
of
test
results
on
test
grid,
I
think
at
least
at
the
very
beginning,
we're
going
to
aim
for
getting
regular
results
on
test
grid.
B
A
So
so,
for
the
day
to
day
test
or
the
checking
PR
based
test,
I
completely
agree,
I
think
there's
going
to
be
valuable
and
that
before
any
release
of
kubernetes
comes
out,
let's,
let's
say
the
major
releases
three
times
a
year
that
do
run
the
entire
test
suite
on
Azure
as
well
and
I
think
this
is
probably
a
need
for
that
to
catch
regressions.
But
my
personal
thing.
E
The
second
one
will
be
to
have
at
least
the
some
of
the
tests
being
executed
them
as
part
of
jobs
by
prowl
in
a
blocking
fashion,
so
that
when
people
submit
PRS,
which
are
going
to
who
cause
I,
show
some
windows,
which
happens
all
the
time,
because,
let's
face
it
most
of
the
people
in
that
you
don't
miss
Oswald,
including
Cogan.
If
you
come
from
a
from
a
Linux
background
er.
So
it's
not
part
of
the
common
mindset.
E
For
example,
saying:
hey,
I'm
user
I
got
a
page,
for
example,
that
user
uses
and
also
specific
and
Linux
signals.
Are
they
going
to
break
on
Windows?
Yes,
now,
ok,
so
in
the
moment
in
which
they
see
the
errors
coming
up
directly
on
the
on
github
and
and
the
patch
not
being
able
to
merge.
Because,
of
course,
that's
the
test
and
already
there
we
will
be
able
to
interact
with
them
and
they
will
be
able
to
submit
an
updated
PRF
before
it
managers.
E
We
were
discussing
this,
and
if
you
do
that
with
desk,
it
approach
you're,
basically
going
to
be
the
watcher
on
the
wall,
meaning
that
you
always
have
to
check
that
something
breaks
and
then
going
and
change
that
the
people
and
submitted
the
page
or
in
general,
the
communities
and
see,
if
the
case
of
Roberti
or
just
submitting
another
patch
to
fix
it.
Ok,
so
that's
good
going
to
be
one
of
the
definite
end
goals
that
we
third
thing
and
also
very
important.
E
A
Yeah
I
think
that
putting
a
very
simplified
PowerPoint,
deck
or
Google
presentation
ahead
of
the
sig
testing
is
gonna
kind
of
frame
this
talking
about
the
milestones
or
goes
with
each.
If
you
guys,
wanna
I
can
kind
of
help
with
that
as
well,
but
I
think
it's
gonna
help
a
lot
frame.
The
discussion
we
sig
test
them.
C
B
A
A
All
right,
Patrick,
let's
talk
about
group,
managed
accounts
I.
You
know
at
the
high
level
I
fully
support
up
top
here
to
make
a
change.
I
think
this
is
a
an
important
part
of
running
Enterprise
type
workloads
on
kubernetes
on
Windows,
so
I
think
that
you
know
whatever
we
need
to
to
get
through
all
I
think
it's
it's
a
very
important
change.
B
Okay,
yes,
I'm
trying
something
a
little
bit
different
here.
So
there's
this
new
tool
that
I've
seen
some
other
people
are
using
called
hack
MD
that
works
sort
of
like
Google
Docs,
except
it's
actually
got
markdown
behind
the
scenes,
which
is
good
because
I
can
go
ahead
and
eventually
submit
that
as
a
proposal
over
on
the
kubernetes
community
repo,
where
we
put
all
the
design
Doc's.
So
that
way,
I'm
not
having
attempted
to
translate
things
later.
B
B
So
there's
a
bit
of
a
background
here
that
most
Windows
users
that
have
tried
this
have
already
run
into,
and
that's
if
you
were
using
docker
or
docker
swarm.
One
of
the
things
that
you
had
to
do
was
create
this
JSON
file
and
put
it
on
all
of
your
nodes
that
wasn't
intended
to
be.
You
know
the
final
way
that
this
feature
was
used.
B
Don't
believe
that
we
actually
need
to
have
that
file
at
all
and
if
kubernetes
is
able
to
consume
either
configuration
maps
or
secrets
on
the
host,
then
what
we
could
do
is
just
pass
that
into
the
container
D
API
as
the
container
is
actually
being
created,
and
so
what
that
would
mean
is
that
you
know
there's.
If
you
look
at
the
section
called
parts
of
a
solution,
there's
basically
two
different
workflows
that
need
to
happen.
B
C
B
A
I'm
totally
on
board
with
this
approach,
it
makes
it
very
simple
for
a
developer
to
consume.
It's
just
products,
ml
specification
for
them.
So
besides
that
ministry,
this
stuff
of
setting
up
the
account
I
think
there,
the
end
user
experience
is
is
simple.
It
doesn't
look
like
would
be
a
tremendous
amount
of
work
for
us
to
basically
make
the
changes
to
go
from
the
pots
back
down
to
the
instructions
at
the
container
level.
A
B
The
next
thing
I
need
to
do
on
my
side,
is
dig
into
the
source
code
and
confirm
that
we
actually
can
consume
a
secret
from
the
host,
instead
of
only
passing
it
down
to
the
pod.
So
that's
just
some
digging
that
I
need
to
do
and
I'll
probably
see
if
I
can
get
there
and
over
over
here
on
my
engineering
team
to
look
at
that
with
me,
because
I
seem
to
make
sure
that's
feasible,
but
maintain
I'm.
B
A
It
might
be
the
case
that
not
everybody
has
had
a
chance
to
read
it.
I
posted
I
know
you
posted
it
on
the
sick
windows
I,
possibly
on
the
chart
as
well.
So
folks,
you
know
folks
from
cloud
base.
Bob
I
love
for
all
of
you
guys
to
have
a
chance
to
look
at
this
Peter
as
well.
Go
ahead
and
read
through
this
and
see
if
this
makes
sense
to
you
guys
and
then
AVS
will
start
moving
down,
make
it
into
proposal.