►
From YouTube: Kubernetes SIG Windows 20200714
Description
Kubernetes SIG Windows 20200714
A
A
I
know
we're
kind
of
marching
towards
a
set
of
deadlines
here
and-
and
hopefully
all
of
you
are-
are
tracking
well
on
that
we'll
try
to
kind
of
take
a
lil
bit
of
stock
of
of
where
we
are
as
well
here
so
I
know
that
some
of
you
are
already
typing
up
furiously
right
now,
adding
some
notes
to
our
agenda
today.
Thank
you
for
doing
that.
I
really
appreciate
that.
So
let's
take
a
look
at
where
we
are
here
in
a
second.
Let
me
show
my
screen.
A
All
right,
you
should
all
be
seeing
my
screen
now.
Okay,
so
the
first
thing
I
wanted
to
take
a
look
at
this
to
date.
So
so
the
current
dates
have
us
on
July
16th,
which
is
in
two
days,
is
the
docs.
Mobic
must
be
completed
and
reviewed
so
I
know
we
have
some
talks
that
James
is
driving.
We'll
have
some
dogs
that
Moz
is
driving.
A
If
there
are
any
folks
that
that
are
that
are
basically
leases
reviews,
please
go
ahead
and
review
those
I
know.
I
have
I,
have
one
of
them
for
continuity.
That
I
need
to
take
a
look
at
and
I
had
already
taken
a
look
at
the
networking
one
that
David
had
created,
but
please
flag
a
flag
them.
If
there's
any
dogs
that
that
focusing
to
review
and
they're,
not
in
our
agenda
for
today,
all
right,
so
let's
go
to
a
notes
really
quickly.
So
I
wanted
to.
A
Excel
the
changes
here
so
so.
Last
week
we
talked
earlier
about
the
cherry-picks
they're
all
approved
now,
so
we
did
the
had
a
conversation
they
released
him
all
the
cherry-picks
are
now
in
all
all
four
of
them.
One
of
them
was
Claudio
by
Claudio
and
the
rest
of
them
were
the
cherry-picks
that
basically
fix
certain
issues
downstream.
So
done
cool
talk
a
little
bit
about
version
1.19,
where
we
are
so
in
point
slices.
David.
Are
you
on
sorry
I'm
sure
my
scan
I
can't
see
everybody.
B
So
for
the
talks
I
think
we
were
discussing
marking
this
as
supported
and
documenting.
It
has
supported
once
we
affirmed
that
it
it's
been
validated
and
it
works.
Someone
knows
I
think
we're
still
waiting
on
those
test
results
to
come
back,
I,
don't
know
if
you
give
there's
only
two
days
for
the
dogs
to
be
completed.
C
A
A
Agree
with
you
David,
since
the
Linux
ones
already
don't
call
out
saying
this
doesn't
work
on
Windows,
then
we're
fine,
so
I
think
it's
the
first
validation
I
mean
the
problem
will
only
occur.
If
we
try
to
validate
and
it
doesn't
work,
then
we
need
to
call
it
out
that
saying
this
is
the
no
charlie
yeah.
B
A
Thank
you
thank
you
and
David
I,
don't
know
if,
if
Mars
reached
out
but
who,
like
one
or
two
slides
on
the
end,
point
slices
and
some
other
networking,
the
DSR
ODF,
these
are
a
thing.
It's
called
BSR.
Yes,
are
so
cows,
eyes
and
DSR
and
end
point
slices
for
for
the
cubicle
presentation.
If
you
can
provide
those
that'd
be
great,
like
yeah.
D
D
So
I
looked
at
the
documentation,
it's
it's
looking
pretty
good.
There
was
one
thing
that
I
think
James
is
going
to
update
we're
the
documentation
on
how
how
to
create
a
package
for
dinner,
tea
and
stall.
It
I
don't
think
it's
part
of
that
documentation
do
it
that
is
documented.
It
just
should
be
either
linked
or
like
we
copied
here.
I.
Think
James
is
gonna
work
on
that.
The
other
thing
I
was
looking
at
is
like
how
much
information
we
need
to
provide
from
container
D
I
think
we
have
enough
information.
D
E
So
I
started
to
the
docks
for
how
to
configure
it
and
install
it.
It's
taking
me
a
little
bit
longer
than
I
thought,
and
so
my
question
is:
do
we
need
to
have
those
in
to
this
PR
and
will
that
affect
because
it's
only
two
days
away
getting
approvals
and
tech
reviews
if
I
have
it
in
by
the
end
of
day
today
or
what
okay.
A
If
you
have,
by
the
end
of
the
day
today,
should
be
fine,
I
think
you
know
the
when
they
say
dogs
must
be
completed
and
reviewed.
That's
that's!
The
completion
date
from
the
sea
sea
Krait
I
mean
the
dogs
like.
As
long
as
we
agree
that
everything
is
good
by
July
16th.
Then
then,
the
rest
of
the
team
from
the
docks
sig
is
gonna,
go
look
at
them.
So
as
long
as
you
have
them
ready
and
we
you
Mema's,
everybody
else
agrees
that
they're
good
and
we're
fine.
Okay,.
E
A
You
know
what
they
added.
All
these
moderation
features
give
me
a
second
James.
The
other
moderation
features
because
they
had
trolls
as
a
uric
co-hosts.
Now,
so
you
should
be
able
to
share
yeah.
There
was
a
troll
that
showed
up
in
one
of
some
other
community
meetings
and
started.
You
know
sharing
things
that
were
inappropriate
and
so
the
kubernetes
country,
with
our
team,
disabled
machete
for
anybody,
that's
not
the
host.
E
So
I
was
so.
This
is
the
section
for
installing
and
configuring
container
D
for
Ubuntu
and
CentOS,
and
so
he's
gonna.
Add
another
tab
here
for
Windows,
perfect
on
and
I
was
gonna
base
it
off
of
how
we
install
it
for
a
KS
engine,
which
is
what
we're
using
for
the
upstream
tests.
So
I've
started
that,
but
it's
just
taking
a
little
bit
longer,
because
I
can't
quite
copy/paste
this
stuff
out
of
a
kiss
engine.
So
so.
A
E
A
D
A
D
B
A
We
should
be
able
to
cherry
picking,
bug
fixes
so
I
mean
this
is
in
the
stabilization
period
right
now.
So,
if
there's,
if
you
guys,
try
to
make
if
anybody
tries
to
make
bug,
fixes
and
they're
not
getting
approved
in
time,
let
me
know
part
of
the
process.
Is
you
know,
attending
release
team
meetings,
so
they
can
actually
flag
it
and
say
this
is
important.
We
need
to
fix
it
in
is
to
going
for
sick
windows.
So
if
you
have
such
a
bug
fix
that's
found
that
we
need
to
check
in
after
code.
A
Freeze,
please
ping
me
privately
on
slack.
We
will
attend
sick
release,
meetings
together
and
try
to
get
them
approved.
You
have
to
make
basically
make
a
business
case
why
they
should
make
B
Cohen
alright.
Next
one
is
2004
update
the
Adelina.
Did
you
add
that
or
was
it
someone
from
microsoft,
test
Crete
for
Windows,
Server
2000
for.
F
E
It
was
me
for
this
one
James,
okay,
so
I've
added
the
test
grid,
there's
about
eight
failing
tests
and
working
with
that
Alena
and
Claude,
who
we
identified,
that
it
was
just
because
the
images
were
not
2004,
and
so
we
I
think
Claude
who
pushed
the
last
couple
that
needed
the
2004
images
this
morning,
and
so
hopefully
tomorrow
will
have
a
test
pass
on
that
and
then
there's
one
for
the
authenticated
image
that
claudia
who
had
said
there's
a
workaround.
So
I'm
gonna
be
working
on
that,
probably
tomorrow
after
I
finished
the
docs.
So
no.
A
E
C
A
C
So
yeah
this
is
continuing,
so
we
have
all
the
combining
support
for
some
of
the
more
tricky
stuff
that
partial
does
not
surface
Cali
added,
quite
awesome
job
on
that.
So
all
that
is
in
the
remaining
portion.
There
is
to
just
bump
the
API
version
from
v1
alpha,
2,
V,
1,
beta
and
and
like
be
applying
to
tweak
one
API
around
the
page.
It's
free
support
that
Callea
worked
on
to
be
exposed
through
the
v1
beta
API,
said
very.
A
C
Think
from
as
you
like,
both
Azure
and
UCPD
have
been
testing
it
for
quite
a
while
and
I
think
they
haven't
run
into
any
major
issues.
So
the
the
main
reason
we
are
doing
this
light
API
tweak
is
to
kind
of
make
it
more
generic.
It's
not
just
something
that
satisfies
gcpd
CBDs
requirements,
but
kind
of
make
it
broader.
C
Right
I
mean
the
overall
CSI
support,
so
things
that
won't
make
it
in
is
like
we
wanted
to
do
a
lot
of
like
fluency
ICD
pipelines
for
CSI
support
across
all
the
all
the
CSI
sidecars.
That's
not
going
to
happen
for
beta
and
that'll
be
mga
mainly.
This
affects
the
no
driver,
registrar
sidecar.
So
all
the
packaging
is
done,
but
CI
aspect
of
that'll
not
be
ready,
but
I
think
we
all
decided
that
that's
probably
okay,.
C
Yeah
yeah
we're
also
getting
some
requests
for
other
things
like
hey.
Can
we
get
like
this
API
or
that
API
from
other
CSI
developers,
which
is
nice,
which
shows
that
people
are
interested
but
again,
like
they'll,
probably
take
some
of
those
requests
more
into
GA,
going
to
get
the
final
one?
Okay
play
God.
C
Can
you
rolling
up
great
the
change
or
is
there
a
breaking
change?
So
this
is
not
going
to
be.
A
breaking.
Change
basically
will
have
just
like
kubernetes
gonna
have
like
all
the
APS
will
be
version,
so
the
order
API
will
continue
to
be
supported
and
that'll
be
exposed
over
a
certain
name,
pipe
saying:
V
1,
alpha
1,
the
new,
the
new
API
is,
will
be
exposed
over
its
own
API
pipe,
which
would
be
B
1
beta,
1
yeah
list
bash
file.
Things
like
that.
So
the
idea
is.
C
We
continue
to
support
multiple
API
versions
with
the
same
proxy
binary
and
we
all
have
an
application
schedule
similar
to
kubernetes,
which
is
b
bx.
We
announced
we
are
going
to
abrogate
something
if
it's
beta
or
GA
for
alpha.
There
are
no
guarantees
and
then,
after
we
declare
it,
we
go
and
remove
it
at
some
point.
A
All
right,
the
next
topic
is
topics
on
storage.
The
first
one
is:
we
actually
had
a
team
of
folks
to
try
out
Valero,
which
is
a
backup
and
recovery
tool
and
tried
out
with
Windows
containers
and
have
both
good
and
bad
data
and
our
kind
of
outlined
in
here
so
for
stateless
apps.
It
works
100%.
So
that's
great
for
non
stateless
apps.
If
there's
persistent
volumes,
for
example,
vSphere
volume
or
DCP
disk
or
ISO
file
or
Azure
disk,
it
doesn't
work
and
just
couple
of
reasons
behind
it.
A
Valera
uses
a
technology
core
astok
for
basically
being
able
to
recycle
pod
and
use
host
host
path
to
extract
all
the
data.
That's
basically
attached
to
a
kubernetes
container
in
a
pod,
and
the
problem
is
that
you
know
resting,
obviously
doesn't
understand
windows
very
well.
So
my
team
went
ahead
and
rebuild
a
drastic
for
Windows
and
they
basically
built
some
Bolero
containers
for
Windows
and
when
rustic
basically
mounted
the
volumes,
every
volume
shows
have
been
a
different
drive
letter.
But
within
the
pod
you
don't
have
access
to
that
drive.
A
It
basically
gets
mounted
on
the
windows
host
as
a
different
drive,
and
you
have
no
access
of
that,
so
unless
we
actually
go
and
modify
rustic
to
actually
work
for
Windows.
So
it
understands
that
when
you
want
something
you
have
to
create
a
first,
you
might
have
to
have
elevated
permissions
to
be
able
to
mount
a
volume
by
basically
mount
the
extra
drive
into
the
location.
Where
you
are
so
you
can
ingest
the
data.
This
won't
work.
A
few
folks
that
I
have
on
this
they'll
they'll,
look
into
the
little
bit
more
and
see.
A
If
there's
anything
smart,
you
can
do.
The
second
thing
that
we
could
do
is
around
CSI
there's
this
effort
around
creating
CSI
snapshot.
This
is
a
snapshot
API
that
that
went
to
beta
with
118
and
once
we
actually
get,
we
might
be
able
to
try
CSI,
plus
windows,
plus
a
snapshot
capability
as
long
as
a
provider
implemented
it,
and
that
could
offer
us
a
seamless
way
to
be
able
to
back
up
and
restore
windows
containers
with
kubernetes.
So
deep
since
you're
like
what
really
and
then
1:19
I,
don't
want
to
kind
of
bring
this
up.
A
C
F
C
A
C
A
We
just
need
to
have
someone
just
try
it
out,
so
I
have
a
couple
of
folks
from
the
Valero
team.
If
you
all
can
work
together,
I
think
that
you
know,
like
you
having
a
cluster,
we
CSI
I've
been
running
everything
ready
and
they
can
come
in
bring
bolero.
You
guys
can
pair
a
little
bit
I
think
we
should
be
able
to
get
it
up
and
running
and
they
can
document
it.
A
C
C
A
A
D
There
was
an
issue
we're
we
were
not
able
to
access
the
SMB
share
on
the
domain
and
that
was
causing
the
the
yeah
the
credentials
not
getting
passed.
So
when
we
looked
at
it
the
way
container
D
access
the
network
compartment
in
the
u.s.
versus
docker,
the
code
pads
a
little
bit
different
and
there
is
an
OS
bug
there,
where
we
need
to
fix
how
you
know
how
this
is
gonna
get
X.
You
know
for
container
D
how
this
is
going
to
work
so
so
I
think.
D
The
reason
we
wanted
to
bring
it
up
is
Craig,
James
and
I
thought
to
bring
that
up,
because
it
feels
like
if
someone
is
using
it
for
GM
sa
this
way.
There
is
that
limitation.
It's
not
fully
covered
in
the
upstream
test,
so
we
probably
need
to
update
the
upstream
tests
for
GM
sa
duduk
and
some
kind
of
SMB
share
and
the
other
update
is
since
it
is
an
OS
change.
We
are
trying
to
fix
it
as
soon
as
possible
with
you
know,
in
2004,
and
then
back
porting
it
to
rs5
in
1981
and
1982.
A
Just
wanted
to
bring
it
up
to
everyone.
Good
question
like
will
like,
for
example,
some
be
sure
student
work,
but,
for
example,
if
you
have
a
database
and
you
are
caused
it
to
an
Active
Directory
account
with
that
work,
this
is
specific
only
to
SMB,
shares
or
is
or
gonna
have
an
identity
issues
are
on
other
areas
as
well.
That
might.