►
From YouTube: Kubernetes SIG Windows 20200121
Description
Kubernetes SIG Windows 20200121
A
Alright,
hello,
everybody
and
welcome
to
another
sig
windows:
Meetup,
it's
the
21st
of
January
and,
as
always,
is
a
recorded
meeting.
So
please
adhere
to
the
CN
CF
Code
of
Conduct
how
everybody
had
a
good
MLK
holiday
for
the
ones
are
US
based
and
you're
ready
to
tackle
a
shortened
week.
We
have
a
pretty
light
agenda
today.
A
So
have
a
few
caps
going
on
so
there's
a
Cuba
diem
cap
and
we'll
have
gab
talk
about
that
and
leave
Ben
started
his
multi
week
vacation
already,
and
then
we
also
have
the
GMs
a
cap
that
we
want
to
talk
about
and
see
if
we're
gonna
believe
dip
may
have
already
updated.
So
we'll
talk
about
that
as
well
and.
A
B
A
To
actually
improve
it,
it
seems
unclear
so
as
soon
as
team
Sinclair
approves
it,
then
that's
the
last
step,
so
it'll
give
the
thumbs
up.
I
think
it
should
be
good
to
go.
I
don't
know,
Tim
is
pretty
busy
if
he
doesn't
reprise
in
the
next
couple
of
days,
I'll
ping
him,
since
we
don't
want
to
go
all
the
way
up
to
the
26th
or
2028
all
right.
Well,
thank
you.
Both
human
ban
for
kind
of
shepherding
that
and
answering
all
of
Loomis
questions
and
making
all
the
necessary
changes.
It
looks
like
you
know.
A
Technically,
you
haven't
changed.
Anything
is
just
some
at
the
high
level.
It
looks
like
they're,
okay
with
that
yeah
all
right.
Thank
you,
the
and
I'm
assuming
cab.
You
guys
are
already
you
know
on
the
implementation.
You
guys
are
going
to
start
looking
into
some
test
Suites
and
automated
tests
that
we
need
to
add
and
all
the
necessary
stuff
to
to
get
all
the
way
to
better
right.
Yeah.
B
A
C
Overall
I
think
we
do
not
have
any
concerns
with
going
to
GA
and
I
mark
that
as
the
goal
in
the
enhancement
issue
for
1:18,
the
one
point
I
think
that's
worth
considering
is
the
only
concern
or
the
only
thought
I
can
think
of
is
like
do
we
want
the
continuity
stack
to
be
ready
and
also
support,
GM
si
before
we
say
that
this
is
GA.
That
was
the
only
thing
I
was
kind
of
thinking
about.
C
Up
to
whatever
you
know,
the
community
decides
so
right
now
the
stack
is
obviously
only
interacting
with
the
topper
engine
and
that's
where
GM
si
is
enabled
it's
someone.
Where
do
you
think
it's
GA
and
then
use
a
container
D
stack
that
feature
may
not
be
available
because
of
you
know
that
the
testing
and
all
that
has
not
been
set
up
with
the
container
to
be
based
set
up.
So
that's
the
only
concern
I
had
basically
any
thoughts.
I
mean.
D
It's
a
from
an
organizational
perspective,
I
think
that's!
Okay,
because
we're
still
planning
to
move
container
D
through,
like
the
alpha
beta,
stable,
designation,
and
you
know
we
can
basically
say
what's
supported-
did
not
up
until
that's
done,
because
I
still
don't
have
an
ETA
on
when
that
work
is
going
to
be
available
in
container
D.
C
C
D
C
A
D
E
A
A
Seconds,
but
you
got
the
gist
of
what
you
mentioned
alright
couple
of
quick
housekeeping,
so
we
submitted
for
two
presentations
for
cubic
on
Amsterdam.
One
presentation
will
be
talked
a
little
bit
about
you
know.
Sick
window
is
a
standard
where
we
are
and
investments
were
making
and
the
second
one
is
a
container
D
advance.
A
A
The
last
thing
is
last
week,
or
was
it
two
weeks
ago
briefly
talked
about
winds
and
some
of
the
security
implications
there
gab
I,
don't
know
if
you
guys,
if
you're
I
know
been
had
a
lot
of
things
on
this
plate
last
week.
Have
you
guys
talked
with
seek
security
about
that
and
kind
of
two
areas,
one
to
make
sure
that
this
is
a
key
part
of
the
next
contest
that
they
don't
kubernetes
or
for
them
to
review
it?
We.
B
A
Them
reviewer
in
the
cab,
because
that
makes
it
a
little
bit
more
formal
for
them
to
review
it
and
then
copying
them
in
the
slide
channel
and
I'm
raising
that,
because
they
might
need
more
than
a
couple
of
days
to
review
it.
So
in
the
cap
to
make
them
an
approver,
makes
them
very
viewer
and
copying
them
sure.
D
A
B
F
A
F
F
I
know,
people
are
familiar
with
what
ple
G
is
or
means,
but
a
it's
a
reminiscent
of
this
issue
from
you
know
a
few
years
ago
that
kubernetes
investigated
anyway,
the
fix
for
us
was
to
change
the
rate
that
we're
creating
pods.
So
when
we
went
from
two
pods,
a
second
to
one
pod,
every
three
seconds
issue
went
away.
We
were
able
to
get
up
to
about
a
hundred
pods
on
our
eight
core
windows
node.
F
So
we'll
dig
in
a
little
more
to
the
right
problem
later
on,
but
I
thought
I'd
just
share
that
and
kind
of
in
case.
Anyone
else
is
running
a
lot
of
pods
really
fast.
We
are
seeing
this
failure.
I
think
my
suspicion
is
that
it's
related
to
docker
and
not
to
the
cube
list,
but
not
totally
sure.
Yet.
Okay,.
F
F
A
F
G
F
D
Yeah
I've
got
a
different
mic
over
here.
Apparently
the
one
in
my
laptop
decided
to
die
good
good,
good
news.
So
what
I've
seen
before
so
that
you
get
two
PL
eg
failure
after
you've
had
multiple
docker
api's
failed
to
respond
within
a
certain
time
period,
and
so
I
think
there
is
still
some
weird
locking
behavior
in
there
somewhere.
D
Mark
and
I
have
been
looking
at
something
over
on
the
met
on
the
cubed
stat
summary,
basically,
where
it
gets
the
pod
usage
data
and
we're
seeing
similar
things
there,
where
once
it
gets
above
a
certain
number
of
pods,
even
though
the
seat
there's
plenty
of
free
CPU.
Some
of
those
api's
go
from
responding
in
less
than
a
second
to
tens
of
seconds,
and
so
we're
not
sure
what
the
case
is
there,
but
it
could
be
related.
D
The
other
thing
I
have
seen
is
that
once
the
CPU
usage
gets
high
enough
on
a
node,
I
have
seen
that
failure,
starting
or
stopping
pods,
and
it
seems
like
it
seems
like
the
context
in
kubernetes-
is
expected
in
docker
to
respond
faster
than
it
can
under
load
and
so
I
don't
know.
If
that's
something
that
we
can
tune
in
the
cubelet
on
Windows
and
like
raise
that
context,
timeout
or
not,
but
I,
think
that's
probably
within
the
realm
of
things,
we
should
be
looking
out
there.
A
D
D
Yeah
yeah
I'm
gonna
update
the
cap,
and
you
know
we're
gonna
continue
working
on
that,
but
it's
at
a
slightly
lower
priority
behind
some
of
the
other
active
issues
like
the
one
like
the
one
I
mentioned
around
the
metrics
and
trying
to
get
a
couple
rough
ends
working
better
on
the
current
releases
before
we
migrate
over,
but
I'll
get
the
cap
updated.
So
that
way,
if,
if
things
are
looking
good,
we
can
declare
it
alpha.
18.
A
Okay,
the
last
thing
for
me
is
that
sonobuoy
team
worked
on
creating
the
ability
to
have
sonobuoy
plug-ins
for
Windows,
so
they
have
a
demo
implementation
now,
as
well
as
a
sample
plugin
that
John
created,
so
that
he
can
basically
demonstrate
what
that
would
look
like.
So
that
kind
of
paves
the
road
for
us
to
start
adding
conformance
tests
in
the
that
will
be
run
by
sonobuoy
as
part
of
the
bigger
suite
that
certifies
and
validates
at
kubernetes
distribution
like,
like
all
the
public
cloud
providers,
have
an
open
shift
as
well
as
VMware
PKS.
A
So
the
next
step
is
I
call
him
in
contact
with
Adelina.
So
another
note
for
the
leanness
on
the
season,
so
Adelina
I,
guess
I'm.
Assuming
that
you
and
John
will
work
and
figure
out.
You
know
what
the
best
path
and
it
is
and
maybe
see
if
there
are
any
Windows
tests
that
we
have.
That
would
be
good
candidates
for
the
conformance
suite
I
know
that's
a
much
bigger
question,
but
at
the
very
least
you
can
start
testing
this
one
Oh
bleep
liking
and
making
sure
that
the
windows
can
be
included.
G
A
I
G
C
A
I
mean
anyone
that
can
kind
of
spend
some
time
on
our
backlog
and
identify
any
items
that
we
may
have
missed
and
and
bring
them
up
to
light
and
say:
hey
I
think
this
is
important.
Let's,
let's
take
a
look
at
that
pressure
me
and
Patrick,
and
Ben
and
D
would
appreciate
that.
There's
always
things
we
miss.
So
if
you
have
some
time
to
go
through
some
of
those
tickets
and
see
hey,
this
looks
like
short
fix
low-hanging
fruit,
but
make
a
big
impact
would
love
to
see
gotta
help.
Okay,.
D
You
know
and
I
dropped
I've
got
the
github
project
that
I
started
on
there
and
it
would
be
great
if
we
could,
you
know,
maybe
start
bucket
izing
stuff
there.
The
other
thing
I
haven't
looked
at
very
closely
is
there
is
some
guidance
around
how
labels
and
priorities
should
be
set
and
we
haven't
been
applying
that
to
our
backlog
at
all,
so
maybe
with.
Maybe
we
should
sit
down
and
start.
D
A
Absolutely
and
then
on
the
on
the
project
board,
make
sure
you
create
a
new
swim
lane.
You
can
call
it.
You
know
your
your
prioritized
thing
that
way
you
give
a
chance
for
folks
to
review
before
we
just
drop
them
into
the
right
buckets.
That
way,
we'll
give
you
the
option
to
organize
the
things
and
have
a
formal
process
around
that:
okay,.