►
From YouTube: Kubernetes SIG Windows 20210309
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
All
right,
hello,
everybody
and
welcome
to
the
march
9th
2021
iteration
of
the
sig
windows
community
meeting
for
the
kubernetes
project.
As
always,
these
meetings
are
recorded
and
uploaded
to
youtube,
so
please
be
sure
to
adhere
the
cncf
code
of
conduct
all
right.
Let's
kind
of
get
started.
I
wanted
to
start
doing
a
little
bit
better
job
of
making
this
community
accessible
to
new
folks.
A
So
I
think
we'll
just
have
a
standing
invitation
for
anybody
who's
new
here
who
wants
to
to
kind
of
just
raise
your
hand
in
slack
or
just
interrupt
me
here,
and
you
can
introduce
yourself
and
every
couple
of
times
we'll
we
can
go
through
and
do
a
quick
high
level
round
of
intros
and
also
start
going
over
some
of
the
the
future
work.
A
I
think
that
these
future
items
that
we
have
on
the
agenda
here
are
a
little
bit
old
and
outdated,
so
we'll
try
and
update
that
in
the
next
couple
of
in
the
next
step
by
the
by
the
next
meeting
or
so
is
there.
Anybody
here
who
is
kind
of
new
here
wants
to
just
introduce
themselves.
Please
kind
of
go
ahead.
A
All
right
an
interesting
time,
I'll
get
going
and
start
with
the
announcements.
First
announcement
is
the
annual
report,
so
we
kind
of
mentioned
this
last
time,
but
I
don't
think
I
did
a
good
job
of
explaining
what
the
purpose
of
the
annual
report
was.
So
I
wanted
to
kind
of
do
that
here
previously,
the
each
sig
for
the
in
kubernetes
project
was
required
to
join
a
the
community
meeting
either
three
times
a
year
or
four
times
a
year
and
kind
of
gave.
A
A
sig
update,
the
I
think
steering
committee
and
the
and
the
contrabex
team
kind
of
had
decided
that
that
was
a
little
bit
just
wasn't
kind
of
giving
the
rights
it
wasn't
attended
as
widely
as
people
wanted
to
and
also
they
wanted
to
make
kind
of
those
updates
more
asynchronous.
So
the
annual
report
is
kind
of
the
a
way
of
also
capturing
some
of
those
guidelines,
so
new
contributors
can
kind
of
come
in
well.
A
New
and
existing
contributors
can
kind
of
come
in
and
take
a
look
at
kind
of
what
the
current
initiatives
are
with
the
sig
and
see
where
they
can
help
out
and
also
just
get
an
idea
of.
What's
going
on
here,
so
jay
started
a
pr
for
this
last
week.
It's
linked
here
I'll
move
that
up
in
a
few
minutes,
so
anybody
who's
kind
of
interested
feel
free
to
comment
on
the
annual
report.
A
Pr,
I
think
we
have
a
couple
of
weeks
before
that
should
get
merged
and
that'll
just
show
up
kind
of
next
to
the
sig
in
the
sig
folder,
so
folks
are
searching
for
sig
windows.
They'll
see
that
next
to
the
link
to
the
meeting
notes
and
everything,
does
anybody
have
any
questions
or
comments
on
that.
A
All
right,
I
don't
see
anything
so
I'll
move
on
next
big
announcement
is
the
20
the
121
code
freeze
is
today
so
we've
gotten
a
bunch
of
prs
merged.
In
the
last
week,
I've
been
trying
to
keep
track
of
all
of
the
pr's
that
we
do
want
to
get
merged.
If
anybody
has
any
prs
that
I
that
I've
missed
here,
please
add
them
to
this
list
myself
and
and
everybody
will
try
and
ping
all
the
reviewers
to
get
these
reviewed.
A
I
believe
most
of
these
are
waiting
for
external
or
from
reviews
from
other
sigs,
so
please
either
just
reach
out
to
me,
or
anybody
in
slack
or
or
add
edit
to
here
and
we'll
try
and
ping
these
and
get
these
in
today
now,
for,
I
believe,
there's
guidelines
for
what
canon
cannot
merge
after
code
freeze,
but
in
general,
unless
it's
kind
of
a
critical
or
bug
or
a
big
usability
issue,
try
to
limit
the
prs
that
get
merged
after
code
freeze
test
freeze
is
next
week
still
so
any
tests
any
prs
for
test
related
cleanup
can
still
go
in,
but
enhancements
and
just
general
code
updates
are
under
going
to
be
in
a
lot
more
scrutiny.
A
For
that
thanks,
I
see
a
couple
things
here.
Is
there
anything
that
anybody
wants
to
call
out
specifically,
I
know
james
like
this,
this
first
one
here,
this
gut
file
system
stats
that
one
I
would
really
like
to
get
in.
I
believe
we're
waiting
on
sig
storage
to
approve
for
that
one.
That
one
makes
the
cubot
logs
kind
of
an
absolute
mess
to
look
at
on
windows.
B
So
I
I
approved
from
sigwin
storage,
but
actually
it's
required
signaled,
which
is
package
cubelet,
yeah
yeah.
I
please
seek
node.
A
I'll
give
a
quick
update
on
the
windows
host
processor
privilege
container
work
most
as
I
was
kind
of
expecting.
Most
of
the
reviews
didn't
come
in
until
yesterday,
there's
not
really
any
contentious
issues.
There
was
one
comment
on
the
pr
related
too
how
we
handle
feature
gates,
because
for
this
one
or
for
this
feature
during
the
prr
review,
the
prr
team
decided
it
would
be
better
to
have
this.
These
fields
not
get
dropped.
A
If
the
feature
flags
were
disabled
and
instead
block
these
privileges
or
host
process
containers
at
admission
time,
so
there
is
a
little
bit
of
a
updates
that
need
to
happen
in
the
pr
for
that
that
I'll,
try
and
address
today.
If
not,
I
will
try
and
file
an
exception
to
get
this
in.
I
think
it
meets
some
of
the
exception
criteria
where
it's
fairly
scoped
and
shouldn't
destabilize
anything
if,
if
it
doesn't
meet
the
deadline
today,
I
think
one
thing
I
did
want
to
call
it
is.
A
I
did
see
a
couple
of
networking
related
fixes.
It
looks
like
they're
getting
added
to
here.
I
have
not
been
doing
a
great
job
filing
following
up
on
the
networking
related
ones,
ja.
I
think
you
opened
some
of
them
or
had
at
least
been
reviewing
them.
Do
you
think
you
could
just
make
sure
that
the
ones
that
we
actually
want
or
feel
like
we'll
merge
for
121,
are
kind
of
captured
in
this
list,
and-
and
let
me
know
if
we
need
to
start
pinging
more
viewers.
C
Yeah
terminating
endpoints,
I
did
not
actually
I
opened
it
and
I
talked
to
andrew
psykim
about
doing
it,
and
then
I
never.
I
think
I
it's
open.
If
somebody
wants
to
dig
into
it
actually
and
yeah,
I
can
I'll
yeah
I'll
do
this
off
offline
mark.
In
fact,
it
might
actually
be
a
good
thing
to
do
for
that.
Onboarding
thing
that
we're
doing
later
on
so
cool
yeah,
yeah
I'll,
look
at
all
those.
A
D
Yes,
so
I
was
going
to
call
out
that
q
proxy
handling
terminating
endpoints.
I
know
that
they're
trying
to
get
that
into
121
before
the
code
freeze,
which
is
today,
so
we
should
ensure
that
whatever
they're
doing
there
doesn't
break
windows
and
I
did
see
david
schwartz
assign
our
cc
sravant
and
keep
monje
from
microsoft
on
it.
So
if
someone
is
someone
from
sig
windows
takes
a
look,
I
think
it'll
be
beneficial
for
us.
We
don't
want
to
be
broken
in
121
on
cue
proxy.
A
Yeah
definitely
all
right
yeah,
I
think
david's
on
the
call.
Well
we'll
try
and
get
the
right
people
to
take
a
look.
C
C
D
A
Yeah,
I
I
saw
a
couple
that
had
work
in
prog
or,
like
maybe
one
or
two
more,
that
had
work
in
progress
peers
opened.
It
didn't
look
like
they
had
a
lot
like.
A
They
were
actually
like
serious
attempts
to
fix
anything
I'll
make
sure
that
those
get
moved
out
of
the
121
milestone
if
I
see
them
I'll
I'll
ping.
The
owners
first,
that
was
another
ask
from
the
release
team,
is
anything
that's
not
critical
or
anything
that
we're
not
really
serious
about
getting
into
the
121
milestone
to
remove
it
from
that
milestone,
just
to
create
less
noise
for
the
that
team
who's.
Looking
at
all
these
issues,.
F
The
cubelet
checking
the
windows
permissions
isn't
particularly
urgent.
It
was
just
it's
been
there
for
a
long
time.
So
all
it
does
is
just
stops
us
getting
that
kind
of
uid.
You
zero
message
when
you
start
kubelet
and
then
tries
to
handle
it
properly,
which
probably
more
important
than
continuity
than
docker,
because
I
think
docker
just
bombs
out.
If
you
don't
have
the
right
permissions,
but
containing
d
actually
tries
to
start
so.
A
C
A
Actually
saw
that
before
as
well,
I
think
that's
a
kind
of
usability
issue
for
people
who
are
looking
at
logs,
so
that
would
be
good
too,
but
thanks
for
calling
that
out
all
right.
One
last
thing
I
wanted
to
call
it
here
is
that
the
coffer
proposals
for
kubecon
north
america
talks
is
open.
There's
a
link
for
submissions
for
that,
it's
open
until
april
11th.
A
If
anybody
is
interested
in
hosting
a
kubecon
talk,
there's
some
some
guidelines
on
kind
of
how
to
submit
a
talk
and
what
will
make
a
good
talk
and
and
all
of
that
stuff
cubecon.
North
america
is
currently
scheduled
to
be
hybrid
in
person
and
virtual
events
later
this
year
in,
I
believe,
los
angeles.
So,
if
you're
interested
in
that
check
it
out,
it's
always
good
to
have
more
kind
of
windows,
representation
in
talks
outside
of
the
maintainer
track.
Talks
too,
I
see
a
couple
of
jay.
A
I
think
you
said
that
you,
this
might
take
about
15
minutes
should
do
you
mind
if
we
run
through
some
of
these
other
yeah,
let's
not
first,
and
then
we
can
kind
of
go
over
that
yeah.
A
I
think
arvind
captured
this
one
jing.
Do
you
want
to
quickly
run
through
some
of
some
of
these.
B
Yes,
so
from
our
test
we
know
his
there
is
a
driver
called
csi
driver
register
fader,
and
that
is
the
error
message.
I'm
just
wondering
I
I
saw
some
old
issue
opened
that
mentioned
this.
The
same
type
of
error
message,
I'm
wondering
whether
anyone
know
what
type
of
failure
it
is,
and
I
also
want
to
point
out
that
this
container
use
nano
server
as
the
base
image.
D
B
A
Much
recollection
of
this
happening
previously,
we
might
need
to
start
a
thread
on
slack
to
take
this
offline
in
general.
The
nano
server
image
has
a
much
smaller
supports,
a
much
smaller
subset
of
the
kind
of
all
of
the
windows
apis
and
how
we've
gotten
around
that
in
the
past
is
the
so,
when
you're
building
images
that
are
targeting
nanoserver
that
need
to
access
the
windows
apis
that
aren't
necessarily
supported
out
of
box
in
nano
server.
A
pretty
common
pattern
is
to
find
the
the
the
user
mode
binaries.
A
That
kind
of
call
in
to
make
those
kernel
calls
from
and
copy
them
out
of
the
server
core
image
into
the
nano
server
image.
It's
possible
that
some
of
those
dependencies
have
changed.
So
the
windows
guarantees
no
compatibility
from
even
os
security
patch
version
to
security,
patch
version
for
the
kernel
mode
calls
and
expect
that
the
patched
kind
of
user
mode
binaries
at
kernel
are
always
in
sync,
and
sometimes
they
have
been
known
to
change
the
kind
of
that
that
those
abi
contracts,
especially
if
there's
a
security
related
fix.
A
If
it's
not
100
percent
repro,
I
would
think
that
nano
server
image
isn't
probably
not
the
cause
of
failure
there.
But
okay.
B
I
can
open
up
this.
Is
we
opened
internally
from
our
testing?
So
I
I
will
open
up
in
open
source
and
I
do
see
the
same
error
message
like
there's
already,
some
open
source
issue
opened
up
with
the
exactly
the
same
error
message,
but
there's
no
conclusion
on
those
issues
too.
G
I
have
a
question
about
this.
If
you
will
so
it
says
that
the
container
exit
unexpectedly,
can
you
actually
start
the
container
itself
doing
a
separate
entry
point?
So,
for
example,
just
have
a
docker
just
pull
the
image
and
try
to
run
the
image
itself
and
do
an
entry
point
cmd
or
something
like
that.
Can
you
run
manually?
The
entry
point
is
my
question:
if
not.
G
Should
work
yeah?
That
would
be
the
first
thing
I
would
try,
because
just
because
you
were
able
to
build
the
image
itself,
it
doesn't
mean
it
works
as
previously
seen
as
previously
seen
by
the
nigerix
failures
and
so
on
and
so
forth.
D
Are
you
sure
that
you're
not
accidentally
landing
on
a
mismatched
post,
because
I
vaguely
remember
seeing
this
sort
of
error
when
the
host
and
container
images
are
mismatched
and
sometimes
the
the
host
container
mismatch
error
message
is
get
gets
butted
and
I've?
Sometimes
you
know,
missed
it
and
later
on,
I
realized.
Okay,
there
was
a
mismatch.
B
G
A
good
point:
can
you
also
link
us
a
link
to
an
image?
I
could
also
take
a
look
if
you
want
me
to.
A
What
irvin
said
is
right:
it's
possible
it'd,
be
good
to
look
at
the
image
manifest
it's
possible
that
the
it's
targeting
like
an
image
from
a
different
like
sac
release
or
windows
version.
G
G
If
there's
an
issue
with
the
image
itself,
there's
a
guide,
which
I
also
followed,
is
pretty
cool:
how
to
determine
what
dependencies
are
required
for
your
application
in
order
to
run
properly
like.
B
A
G
B
Okay,
so
the
second
issue,
like
last
time
I
mentioned,
we
discovered
os-
that
does
not
work
for
socket
file
and
so
right
now
in
during
the
plugging
water,
we
we
change
it
to
use
l
stats
for
windows
and,
at
the
same
time,
I
also
test
os
that
that
always
does
not
work
for
a
singling.
That
is
a
point
to
a
director,
a
directory.
B
B
And
the
last
thing
also,
you
probably
know
for
volume
we
have
amphitheater
volume,
that's
for
like
scratch,
space
you
like
caching
or
logs,
so
I
think
from
implementation,
even
though
it
allowed
the
memory
as
the
medium,
but
it's
actually
using
the
local
disk.
So
basically
memory
as
the
medium
for
emulator
is
not
working,
so
it's
actually
still
using
the
local
disk.
I
want
to
see
whether
anyone
know
to
support
like
similar
like
a
temp
fs
in
windows.
A
Yeah,
I
saw
some
of
that
thread
and
there
are
kind
of
some
third-party
tools
to
help
create
a
like
a
I
think
in
on
windows.
You'll
see
that
more
kind
of
referred
to
as
a
ram
disk
to
create
some
of
those
I've
never
seen
any
really
compelling
use
cases
for
them
on
windows.
I
don't
know
anybody
else
here.
B
A
Oh
lumiere
just
said
that
I
think
we'll
go
back
to
the
os
stat
issue
after
this
one
wraps
up.
Oh.
B
That's
the
handout
okay,
so
because
temp
fs
for
linux
red
is
using
memory,
so
it's
actually
much
faster
than
the
disk.
I
don't
know
for
windows.
There's
no
such
use
case
like
so
memory
could
be
much
faster
in
terms
of
performance.
A
Yeah,
I
I
don't
have
a
lot
of
details
here.
I
just
know
that
in
general,
there's,
I'm
not
sure
if
there's
just
a
lot
of
complications
with
how
they
work
in
windows
or
anything-
I
don't
know
muzz
or
kevin
or
maybe
jocelyn.
Do
you
have
any
comments
for
this?
If
not,
we
can
go
back
to
the
os
dad
issue
and
kind
of
either
create
an
issue
or
bring
this
up
on
slack
again,
the
empty
dirt
sure.
D
Yeah,
I
don't
know
about
ram
desk
on
windows.
G
Yeah,
I
think,
let's
do
this
on
slack,
we
may
be
able
to
get
the
answer
from
some
of
the
folks.
A
All
right
all
right,
lumiere,
sorry,
for
I
didn't
see
that
you
had
to
raise
your
hand.
Do
you
want
to
comment
on
the
previous
item.
H
Yeah
no
problem:
I
just
wanted
to
understand
more
about
the
issue
because
it's
usually
usually
a
good
idea
to
start
logging
tickets
when
we
find
bugs
in
the
standard
library
of
gold-
and
I
myself
have
sent
a
couple
of
fixes
for
windows
in
golang
in
the
past,
so
to
be
clear.
H
B
Yes,
yes,
I
just
created
right
in
my
local
kind
of
my
user
account
just
create
a
normal
directory
and
just
create
a
singling
point
to
that.
Yeah.
Nothing
special.
H
B
So
I
open
up
the
issue
on
the
the
windows
repo
for
os
stats
and
also
I
mentioned
this
issue
together.
Basically,
I
open
up
issue
for
two
two
problems.
One
is
the
circuit
file,
the
other
is
this
link.
H
H
A
The
other
thing
that
might
be
worth
mention
or
trying
is
try
using
the
x
cis
windows
library,
I've
seen
a
couple
of
issues
with
you
know:
file
stuff
that
where
people
have
opened
up
issues
in
against
the
standard,
oi
schooling,
library
and
we're
kind
of
redirected
to
use
the
the
xsys
windows
library,
which
does
a
lot,
I
think
it
just
in
general
makes
more
use
of
windows
specific
like
system
apis,
and
if
that
doesn't
work.
Yes,
we
should
file
an
issue
in
golang
for
this
somewhere.
G
A
G
I
have
a
very
small
question.
I
don't
know
if,
if
you
notice
or
not,
is
the
folder
sim
link
or
sim
link
d,
for
example,
if
you
do
a
deal
list
in
cmd
or
something
you
can
actually
see
that
one
might
be
simply
consuming
d,
because
if
it's
not
assembling
d,
then
it
won't
be
able
to
be
properly
properly
followed.
G
Yeah,
that's
actually
that
at
least
three
of
them,
if
you,
if
you
do,
I'm
calling
slash
question
mark
but
from
what
I
saw,
there's
also
an
issue
in
kubernetes
as
well,
so
it
is
a
fix
for
it,
but
it's
not
related
to
this.
Specifically
at
one
point
a
folder
is
created
after
the
sim
ring
is
created,
and
because
of
that,
it's
the
sim
link
itself
is
not
sibling
d.
It's
just
sim
link
and
because
of
that
it
cannot
be
followed.
G
H
Well,
I
see
that
start
is
doing
get
file
attributes
extended
and
if
that
fails,
it
go
falls
back
to
create
file
yeah
mark.
I
think
that
your
assumption
that
we
might
be
redirected
to
the
extended
you
know
the
goals
extended
standard
library.
H
It
might
be
true,
I'm
going
to
check
this
gene.
Please
make
the
issue
I'm
going
to
log
a
ticket
and
talk
to
the
go
maintainers.
H
A
All
right,
it's
10
a.m,
so
I
think
I'm
going
to
drop
and
head
over
to
the
signate
call
at
this
point
I
think
I'll
hand
it
over
to
jay
to
either
finish
up
the
agenda
or
start
some
of
kind
of
the
after
hours
topics
that
he's
been
starting
so
take
care.
Everybody
I'll
see
you
next
week.
C
Yeah
we
can
end
the
call
and
then
restart
the
other
one
in
a
second.
C
Or
just
we
just
restart
the
recording
is
that
how
it
works?
Yeah,
yeah,
okay,
cool!
Well,
let's
see,
if
anybody's
here
today,
I
think
the
only
person
last
week
was
friedrich,
but
I
don't
know
if
he's
here
today
so.
C
Let's
see,
let's
see
who
drops
and
who
stays,
and
meanwhile
we
can.
I
can
dig
around
and
see
I
kind
of
like
that
thing
that
mark
brought
up
about
the
the
coop
proxy
issue,
because
we
could
look
at
that
issue
and
we
could
get
some
actual
work
done
while
actually
going
through
some
of
the
windows
coupe
proxy.