►
From YouTube: Kubernetes SIG Windows 20170725
Description
Kubernetes SIG Windows 20170725
A
Okay,
guys
I
think
we
can
probably
get
started
now.
My
name
is
Apes
Alton
and
I'm,
filling
in
for
Michael
Michael
from
aprender
Michael
is
out
on
vacation.
He
asked
me
to
run
the
meeting,
so
it
just
asked
me
to
as
per
a
few
things,
and
he
send
me
a
quick
agenda,
but
if
you
have
any
other
questions
or
anything
else
that
you'd
like
to
share
by
all
means,
please
do
so
I
guess
just
for
recording
purposes.
Today
is
Tuesday
July,
25th
2017
and
we're
recording
sick
windows,
so
I
guess
to
get
started.
A
I
know
that
there's
some
work
that's
happening
both
on
a
friend
a
side
of
Microsoft,
side
and
I.
Believe
a
few
of
the
cloud
base
guys
also
making
some
progress
and
networking.
So
just
let's
just
start
with
a
with
some
updates
as
well
whoever's
ready,
maybe
Bob
or
Alex
from
a
trend
edge.
You
guys
can
give
a
quick
update
and
then
we'll
go
around
okay.
B
Hi,
so
we've
begun
in
our
initial
investigation
into
volumes,
config,
meds
and
secrets,
so
those
just
started
but
we're
just
in
the
way,
though,
and
seeing
what's
fleuret
and
hope
to
get
started
implementing
some
of
those
features.
A
All
right,
excellent
I
believe
maybe
Dave
Porter
from
Microsoft.
Was
there
any
thoughts
on
a
possible
demo
for
so
today,
so.
C
First
of
all,
I
didn't
meet
all
of
you
guys
yet
I'm
from
Microsoft
I'm
on
the
azure
container
service
team
and
I'm
an
intern
this
summer
on
the
team
so
I'm
working
on
adding
stacks
to
the
Windows
Server
containers
on
kubernetes.
So
to
give
you
a
quick
update
on
where
I'm
at
I
posted
an
issue
on
the
kubernetes
project,
that's
not
kind
of
detailing
my
plan
here.
C
I
can
post
it
in
the
chat
and
basically
into
summarize
what
I'm
doing
is
currently
there's
a
stat
summary
endpoint
that
metrics
are
read
from
and
the
metrics
are
read
from
a
component
called
P
advisor
that
does
not
support
Windows.
So
what
I'm
doing
is
basically
adding
kind
of
kind
of
fake
C
advisor
in
some
sense
by
reading
per
counter
metrics
directly
from
windows
as
well
as
reading
stats
from
from
doctor
sets
the
other
the
rest,
the
rest
api.
C
C
C
This
is
currently
what
the
what
experience
is
right,
so
this
is
going
to
be
in
the
coolest
component
and
right
now,
if
you
hit
the
stat
endpoint
and
the
steps
endpoint
is
used
by
keep
stir,
for
example,
to
collect
metrics
as
well
as
for
like
auto
scaling
and
other
things,
they
all
call
this
endpoint.
So
if
you
call
this
endpoint
right
now,
it
crashes
the
coolant
and
it
restarts,
but
that's
credit
experience
right
now,
so
what
I've
done
first
of
all,
I
fix
the
the
crashing.
C
So
it's
a
good
start
and
right
now,
if
you
this
is
the
same
endpoint.
If
I
just
do
a
curl
on
it,
you
get
the
CP
metrics.
So
this
is
the
usage
metrics,
for
example,
when
I'm
still
working
on
adding
the
other
other
metrics.
For
the
individual
pots
and
stuff,
but
that's
kind
of
my
current
progress,
so
this
CP
metrics,
for
example,
is
coming
directly
from
counters
and
I'm
reading
that,
because
the
stats
endpoint
guess
that's
both
for
actual
house,
so
that's
the
node
as
well
as
for
an
actual
positive
on
the
on
the
note.
A
Actually
David
I
think
you
might
have
mentioned
it
in
the
ticket,
but
so
a
little
bit
ago,
sometime
last
year,
we
had
tried
to
go
down
this
route
and
we
talked
to
some
of
the
see
advisor
contributors
and
they
had
said
to
look
into
the
CRI.
Instead,
have
you
been
able
to
follow
along
any
of
that
conversation
or
yes,.
C
Mam
so
I've
definitely
been
looking
into
the
season.
Two,
the
Syrah
and
stuff
and
I
talked
a
little
bit
about
that
in
my
issue.
So
the
problem
with
the
CRI
is,
it's
still
little
ways
out,
and
it's
still
in
development
and
the
CRI
simply
they
like
interface
for
the
metrics.
So
the
idea
is
idea.
Basically
right
now
is
that
sentence.
Cr
I
work
for
metrics
are
still
pretty
early.
C
What
I
couldn't
decided
to
do
was
the
first
kind
of
go
down
this
route
and
I
looked
at
that
I
looked
at
kind
of
what
it
would
take
down
the
road
to
implement
to
get
here,
I
working
and
we
can
probably
reuse
most
of
the
code,
because
the
CRI
simply
kind
of
interface.
So,
if
you
you
know,
I
can
read
from
Locker
stats
from
other
things,
so
it
would
just
be
kind
of
implementing
the
interface
for
CRI,
and
a
good
thing
also
is
that
the
CRI
had
implemented.
C
They
talked
about
kind
of
something
called
a
common
set
of
container
minimum
required
container
metrics,
which
is
actually
good
in
the
sense
that
right
now
there's
a
whole
bunch
of
metrics
that
are
being
a
read
and
not
all
of
them
are
actually
being
used
for
for
stuff
like
auto
scaling,
so
they
define
a
minimum
set
and
I've
kind
of
identified.
What
what
that
is
in
terms
of
on
Windows
side
and
I'm,
going
to
be
supporting
that
from
kind
of
a
few
other
things.
So
most
of
the
work
will
already
be
there.