►
From YouTube: Kubernetes SIG Windows 20210706
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
Welcome
everybody
it's
july
6th.
This
is
sig
windows,
weekly
meeting.
This
is
a
cncf
meeting.
So
if
you
have
we
follow
the
code
of
conduct,
you
have
any
questions
or
concerns
you
can
reach
out
to
myself
or
any
of
the
other
leads
and
kubernetes,
and
we
will
address
them
welcome.
A
A
All
right
cool,
so
I
welcome
back
mark
mark,
has
been
on
vacation
for
extended
period
of
time,
so
excited
to
have
him
back
and
yeah
go
ahead.
Mark
too,
say
anything.
A
Excellent
yeah
really
excited
to
have
you
back
so
yeah
thursday
is
code,
freeze
so
make
sure
if
you
have
any
aprs
open
to
get
those
pushed
through.
I
know
there's
a
couple
that
we're
tracking
perry
in
particular
for
the
host
process.
I
think
we
need
to
get
that
in
so
that
we
can
be
able
to
run
someone
eyes.
A
Did
you
just
drop
that
link
in?
I
think.
A
D
A
And
I
saw
you
posted
an
update
on
that,
so
I
we're
still
having
a
conversation
internally
to
see
how
we
can
handle
that.
But
maybe
we
can
chat
about
that
here
in
a
minute
too,.
A
No
okay,
so
I
think
this
is
sort
of
kind
of
announcement.
Also
just
wanted
to
bring
awareness
and
say
thank
you
to
everybody
that
worked
on
it,
but
we
have
our
city
windows,
dev
tools.
Repository
was
created
last
week
and
it's
now
I'll
just
open
it
up
here.
So
I
think
we're
calling
it
a
batteries
included
development
environment
for
kubernetes
and
windows.
A
So
thanks
to
sladen
and
friedrich
for
for
driving
that
and
jay,
it's
really
great
to
see-
and
I
know
that
there's
continued
work
being
done
on
that.
So
does
anybody
want
to
say
anything
you're
going
to
call
or
jay.
F
F
It's
actually
a
bug
with
andrea
in
the
way
that
andrea
deals
with
multiple
mix,
and
so
I'm
going
to
possibly
I've
pushed
some
changes
up
this
week
in
there,
but
I
also
wanted
to
thank
perry
for
being
our
first
customer
and
getting
the
post
process
stuff
tested
on
there,
which
is
really
cool,
but
I
want
to
add
calico
into
it.
F
So,
if
anybody's
looking
for
a
quick
little
project
to
help
us
out,
adding
an
alternate
cni
would
be,
I
think,
relatively
easy
to
do,
because
we
already
have
andrea
working
and-
and
so
it's
like,
there's
a
template
you
could
follow.
If
not
I'm
going
to
add
it.
This
week,
though,.
F
A
Okay,
I'm
not
sure
who
added
this,
but
is
there
oh.
F
I
just
added
that
that's
related
to
what
I
just
mentioned
and
it's
a
kind
of
a
granular
cni
issue.
So
maybe,
since
we
don't
have
a
long
agenda,
we
can
go
to
the
next
item
and
if
folks
want
to
dig
into
how
cni
works
on
the
dev
environments
with
me,
we
can
do
that
after.
A
Okay
sounds
good:
let's
open
this
one
up.
This
is.
A
This
is
the
host
process,
so
harry
you
want
to
go
for
this.
C
Yeah
so
while
I
was
doing
the
testing
of
the
host
process
sort
of
functionality
on
122,
what
I
noticed
was
that
there
was
no
volume
stuff
in
there
yet
so
danny
kindly
went
through
and
started
adding
that,
but
in
when
he
was
doing
it.
What
we
noticed
was
that
it
was
automatically
trying
to
mount
the
etsy
host
file
from
linux
into
windows
and
obviously
couldn't
do
that
because
it
wasn't
under
the
under
the
sim
linked
path.
C
So
what
I
put
in
here
was
basically
just
a
remover
of
some
stuff
that
was
taken
out
last
year.
I
think
basically,
what
it
looks
like
is
we
used
to
always
skip
the
mount
etsy
host
file.
Then
it
was
changed
to
say
if
it's
container
d,
then
we
do
do
it,
but
I
don't
think
it's
actually
doing
anything
for
us
in
windows,
and
so
I
thought
well,
we
just
explicitly
skip
it
because
I
don't
think
we
it
would
help
us,
even
if
we
had
it
mounted
but
yeah.
C
So
I
just
it
was
ageless
sort
of
looking
for
feedback.
I
just
wanted
to
bring
it
up
and
understand
why,
if
ever,
we
would
want
to
mount
the
etsy
host
file
from
linux
on
the
windows
or
whether
we
should
be
mounting
something
to
do
with
the
host
file
or
whether
I
think
james,
you
made
a
suggestion
that
we
added
into
single
file
mapping,
but
I
wasn't
really
sure
if
we
added
it
there,
whether
that
would
break
something
else,
so
I've
kind
of
been.
C
G
C
G
Basically,
kubernetes
and
cubelet
writes
such
a
file
for
every
single
container
and
then
mounts
it
into
the
container
to
say
so
so
yeah
we've
also
added
this
file
into
windows,
because
container
d
is
able
to
mount
single
files
into
container
into
the
containers.
Unlike
darker,
so
we
added
that
just
to
be
more
future
complete
to
say
so.
G
I
mean
that's
what
I
always
thought
that
the
that
particular
face
still
useful
right.
If
I'm
not
mistaken,.
C
So
I
I
was
under
the
impression
that
it
would
only
work
if
it
was
on
host
network,
but
maybe
I'm
wrong,
maybe
I'll
miss
something
because
I
didn't
think
anything
supported
host
networking
at
the
moment
unless,
unless
I'm
wrong.
G
G
B
C
A
Yeah
I
mean,
I
think
the
the
major
point
here
is
that
for
host
process
containers
in
windows,
it
just
doesn't
make
sense
to
have
the
have
this
file
because
it
would
conflict
across
all
the
different
processes
that
are
running
on
the
host,
and
so
I
think
we
need
to
skip
it
here.
I
guess
now
that
I
think
about
it.
Maybe
single
file,
mapping
support,
isn't
exactly
what
we
want
to
skip
it
in,
but
maybe
maybe
something
else
up
here,
but
I
think
we
still
want
to
skip
or
include
that
other
file
into.
B
B
C
I
guess
I
guess
if
it
was
host
network,
then
you
could
just
create
a
sim
link
from
the
pods
etsy
host
to
the
like
the
the
one
that's
on
the
host.
But
then
then
I
don't
the
correct
move
or
wrong,
but
they're,
not
real
containers
in
this.
In
the
same
sense,
are
they
they're
kind
of
processes
on
the
host
that
yep
kind
of
run
in
a
kind
of
sandbox?
C
C
A
C
Okay,
if
anyone
has
any
ideas
on
how
we
could
change
this
before
thursday,
please
feel
free
to
just
throw
it
my
way
and
I
don't
mind
doing
the
typing
work,
but
I
have
it'll
just
save
me
some
time,
but
if
we
don't
know,
then
I'll
carry
on
looking.
A
Yep,
I
think
we
at
least
have
a
path
forward
for.
G
F
I
well
I
mean
yes,
does
anybody
else
have
any
other
stuff
before
I
sort
of
you
know
monopolize
this
this.
This
call.
E
H
D
A
Projected
volume
yeah,
I
don't
have
any
updates
on
the
predictive
volume
thing
I
saw
what
you
had,
but
I
think
there's
still
a
discussion
happening
internally
on
figuring
out.
What
is
the
the
expected
behavior
so.
D
Okay,
yeah
sounds
good
I'll
I'll,
keep
my
eyes
open
for
that,
then
yep
try
to
get
something,
and
just
so
that
we're
clear
james
like
there's
nothing.
I
can
do
now
to
fix
this
right.
I
couldn't
I've
exhausted
all
of
my
attempts
to
sort
of
fix
this
with
what
with
what
I
know
at
the
moment:
yep.
Okay,
all
right
thanks
james
sorry,
jay,
take
over.
F
No
problem
thanks
harvan,
so
the
default
gateway
issue.
What
I
found
was
we
we
so
so
after
setting
some
of
the
interest
stuff
up
in
the
dev
environments,
I
found
that
the
android
agent
was
restarting
and
the
reason
andrea
agent
was
restarting
is
because
it
wants
a
default
gateway
to
already
exist
on
the
node
ip
of
the
kubernetes
node
address.
F
So
so
I
started
looking
at
why
and
it
turns
out
that
I
filed
an
issue
in
upstream
mantra
and
they're
working
on
it.
It
turns
out
that
the
reason
it
wants
that
is
because
it
extends
open.
It
extends
hns.
So
when
it
when
it
goes,
and
it
creates
an
hns
hms
network
or
hcn
network
or
whatever,
whatever
we
call
it
and
it
creates
a
hdm
network
and
through
the
hts
shim
api
and
then
after
it
makes
that
network.
It
then
creates
an
ovs
extension
that
it
gloms
onto
that
network.
F
I
That's
a
question
that
I
might
be
able
to
answer.
Transparent
mode
means
just
it's.
F
I
C
Not
necessarily,
I
think
I
think
you
can
enable
container.
I
think
you
can
enable
the
containers
feature
without
enabling
the
hypervisor,
but
then
what
happens?
Is
you
miss
a
load
of
the
vm
switch
functionality?
I
think
we've
raised
this
right.
I
think
we
put
this
on
by
windows,
containers
that
we
wanted
and
a
separate
api
to
say,
enable
vm
switch
rather
than.
F
Now
let
me
ask
you
a
question:
do
do
we
need
cat?
Do
we
need
then?
Since
so
does
it
do
all
cni's
require
you
to
enable
hyper-v
by
that
api?
Or
is
this
just
some
weird
android
thing,
because
we
make
that
hyper,
because
we
make
that
html
extension,
because
this
hts
hds
gem
api
that
we're
calling
ultimately.
I
F
Okay,
all
right
cool,
so
anyways
that
was
a
little
bit
of
a
sort
of
interesting
detour.
It's
not
really
the
core
problem,
though
the
core
problem
is.
We
need
that
whole
thing,
but
ultimately
what
happens
is
andrea
then
goes
and
it
it
creates
that
transparent
network
and
after
it
creates
a
transparent
network.
F
Then
well
so
what
it
wants
to
do
is
it
wants
to
then
add
a.
It
then
wants
to
configure
ovs
to
have
an
uplink
in
an
open
vswitch
route
so
that
that
open
vswitch
routes,
uplink
is
the
default
gateway.
F
So
there's
a
problem
here,
because
when
we
spin
up
in
our
development
environments,
our
cni
virtualbox
does
a
weird
thing
where
it
creates
a
bridge
and
that
bridge
is
connected
to
the
default
virtualbox
router,
but
it
also
creates
a
private
network
ip
and
then
I
tried
a
few
things
like.
I
tried
to
use
a
public
network,
but
then,
when
windows
server
came
up
and
I
tried
to
make
use
a
public
network
windows,
server
literally
had
no
ip,
and
I
had
no
idea
why.
So
that
didn't
really
work.
F
And
then
I
tried
a
few
different
things
and
nothing
seemed
to
to
make
it
happen.
So
then
I
finally
like
hacked
up
the
intro
source
code
and
I'm
testing
that
now
to
see
if
I
can
just
have
andrea,
put
some
other
default
gateway
in
there.
But
I
guess
my
question
is
yeah.
Go
ahead,
feel
free
to
interrupt
me
because
I'm
kind
of
like.
I
F
Yeah
I
mean,
if
you
have
someone
that
could
look
into
it,
and
just
since
this
is
so
jocelyn
said,
if
you're
not
aware
our
dev
environments
are
literally
you
clone
them
down,
and
you
just
run
make
all
you
don't
have
to
do
anything
else
and
it'll
create
them
all
from
scratch
and
I
think
perry
said
he
got
danny
using
them
too.
So
it's
actually
a
very
very
it's
like
one
step
to
reproduce
it
right.
It's
not
any
anything
complicated
and
I'm
happy
to
tap
on
the
zoom
with
them
as
well.
F
But
I
will
I
will
put
the
details
of
this
in
slack
for
sure
awesome
and
that's
all
I
got
and
then,
since
I
just
got
back
from
ptl,
I'm
just
like
buried
and
stuff.
So
I
for
pairing
this
week.
I
I
do
not
know
if
I
can
drive
pairing
is
anybody
else
want
to
drive
it
or
do
we
want
to
skip
this
week
and
just
start
back
up
next
week?
F
F
A
Okay,
is
there
we
got
two
minutes
left?
Is
there
anything
else
anyone
wanted
to
bring
up
or
mention.
C
I
was
just
just
gonna
say
I
think
I
can.
I
can
check
whether
the
windows,
if
it's
gonna,
be
on
host
network
like
pod
spec
host
network
in
that,
so
I'm
gonna
just
basically
skip
it.
If
windows
host
network
is
selected,
oh
well,
if
it's
host
network
and
it's
windows,
then
explicitly
don't
mount
it
is
that
better
than
just
saying
if
it's
windows.