►
From YouTube: Kubernetes SIG Windows 20220823
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
Hello,
everybody
and
welcome
to
the
august
23rd
iteration
of
the
kubernetes
windows
community
meeting.
As
always,
these
meetings
are
recorded
and
uploaded
to
youtube
so
be
sure
to
adhere
to
the
cncf
code
of
conduct.
Let's
get
started
with
some
announcements
today
should
be
125
release
day.
I
know
yesterday,
all
of
the
different
sub
teams
on
the
release
team
gave
a
go
signal
so
that
they're
just
I
don't
think
that
there's
any
blockers
for
the
release
so
excited
to
see
all
the
work
that
was
done
over
this
release.
A
Does
anybody
else?
Have
any
announcements
they'd
like
to
add
or
anything
I'd
like
to
share.
A
B
A
And
then
I
saw
you
turned
your
camera
on.
So
I
was
like.
Are
you
gonna?
Add
something
yeah
exciting
times
start
planning
for
126
pretty
soon
too
all
right?
Next,
we
can,
if
there's
any
new
contributors
here
we
can
kind
of,
let
them
say
hi.
I
actually
don't
see
anybody
new
on
the
call.
Oh,
do
you
see
a
couple
new
people
so
yeah?
A
All
right
I'll
go
on
to
the
agenda
first
agenda
item
is
something
that
I
added
well.
A
Let
me
see
where
to
start
so
as
if
anybody's
kind
of
been
joined
for
a
while,
we
have
a
github
board
that
we
use
to
track
all
of
the
different
issues
and
pr's
for
sig
windows.
This
was
the
old
style
github
board.
A
The
new
github
projects
have
been
kind
of
out
of
beta
recently,
so
I
was
experimenting
with
some
of
those
and
I
created
a
new
github
project
for
specifically
for
sig
windows
issues.
I
had
one
of
my
kind
of
pet
peeves,
where
the
old
board
kind
of
mixed
the
issues
and
the
prs
in
a
way
that
was
hard
to
kind
of
separate
them,
but
anyway
the
new
github
projects
board.
A
I
don't
know
if
people
are
familiar
with
the
github
projects,
they're
they're,
very
flexible,
you
can
view
them
as
spreadsheets
there's
a
lot
of
filtering
and
it
actually
makes
it
very
easy
to
switch
the
view
to
the
old
boards.
So
I
think
pretty
much
nothing
was
lost,
but
there's
a
lot
of
benefits
here.
A
There
is
one
big
limitation
right
now
and
that's
that
on
these
new
project
boards
through
the
apis,
you
cannot
add
issues
from
other
orgs
to
a
project
board.
So
that
means
you
can't
add
through
automation,
any
of
the
issues
in
the
kubernetes
cigs
repositories,
with
that
I've
got
some
issues
open
and
got
the
ear
of
some
of
the
github
project,
people
to
try
and
get
that
resolved,
but
yeah
if
anybody's
interested
in
helping
out
well,
I
hope
to
start
using
this
at
our
bi-weekly
like
triage
and
backlog
refinement
meeting
but
yeah.
A
Just
let
us
know
if
there's
any
feedback,
I'll
probably
try
and
do
the
same
thing
and
make
either
another
project
border
add
prs
to
this
board
so
that
we
can
start
tracking
them.
A
But
we'll
we'll
see
how
everybody
likes
using
this
yeah
you
can.
You
can
make
different
views
on
the
same
issues
and
save
them,
which
is
also
very
nice.
I
think
it's
a
lot
better
than
using
the
little
text
filter
in
the
old
boards
yep.
So
there's
a
link
to
the
board.
I
can
link
to
the
automation
and
the
scripts
that
are
adding
this.
If
people
are
interested
too
but
yeah,
let
us
know
what
you
think.
B
Yeah
and
we
do
triage
every
other
week
on
thursdays
and
everybody's-
welcome
to
join
us.
We
kind
of
just
go
through
the
issues
and
figure
out.
You
know
something's
the
actual
issue
or
if
we
can
answer
it
right
away
or
we
need
more
investigation
and
that
type
of
thing
so.
A
Yeah
it's
at
the
same
time
as
this
meeting
so
9
30,
pst,
12,
30
est
every
other
thursday.
I
think
there's
a
slack
bot
reminder
too.
That's
all
I've
got
for
that.
Most
people
have
questions
arvind.
D
I
don't
have
anything
to
share
such,
but
I
just
wanted
to
drop
this
in
there.
I
could
show
a
demo
if
people
want
it,
but
I
don't
know
so
what
I've
done
with
the
nerd
log
we
are
now
is,
instead
of
using
a
separate
service
and
part
options,
there
is
a
single
option
that
says
dash
query
and
I,
I
think,
did
I
link
to
the
wrong
pr.
This.
D
D
C
D
So
I
have
I'm
using
the
sig
windows
devtools
environment,
to
bring
up
my
stuff.
So
you
have.
I
have
these
two
instances
here:
there's
one
there's
a
control
plane
and
there's
the
windows
instance.
So
now,
if
I
look
at
the
alpha
command,
I
can
do
something
like.
D
Cube
cto
alpha
node
logs,
I'm
using
a
row
option
and
saying
I
want
a
query.
D
I
just
dropped
a
file
called
test.log
inside
a
test
directory
and
you
know
it'll
give
you
an
output
that
says
no
log
rear
test
because
that's
what's
there,
you
can
also
look
at
the
output
of
all
the
files
that
are
in
that
in
that
location
and
then
you
can
query
any
of
these
files
now.
Alternatively,
if
I
do
something
like
cubelet,
it
will
show
me
like
basically
the
cubelet
logs
for
that
node.
You
can
also
do
container
d
and
it'll.
Show
me,
like
you
know,
continuity
locks.
D
You
are
also
able
to
to
query
the
nodes
directly,
so
I
can
do
something
like
oc.
No.
D
D
So
here
you
can
see
I've.
I've
queried
for
a
service
called
test.
The
the
logic
figures
out
that
this
is
not
a
journal
log,
a
service,
it's
log
into
a
file,
and
it
has
a
heuristics
to
go
and
look
through
all
the
files
and
it'll
and
it'll
show
you
this.
So
I
can
do
the
same
thing
for
windows.
D
Yeah
so,
for
example,
I'm
querying-
I
don't
know,
I
found
this
service
called
microsoft
windows,
ms
dtc,
you
have
no
idea
what
it
does,
but
it
will
show
you
the
logs.
Hopefully,
I've
also
noticed
that
getting
the
logs
from
the
windows
instances
takes
a
little
bit
longer,
and
I
also
notice
that
I've
given
a
30
second
window
for
getting
all
the
logs.
If
the.
If
your
logs
are
really
big
on
windows,
it
gets
truncated
after
a
while
or
you
don't
sometimes
see
the
full
logs.
D
You
can
also
do
like,
for
example,
yeah.
So
if
I
want
to
look
at
the
cubelet
error
error
log
you
can
you
can
see
that
you
can
do
the
same
things
that
you
do
with
on
the
on
the
linux
side,
nice
yeah,
so
all
of
it
is,
is
working
now
I
don't
know
what
the
response
from
tim
and
and
jordan
are
going
to
be.
So
I'm
just
I'm
just
going
to
wait
for
that.
The
only
thing
that
I
I
so
the
way
I've
done
it
is.
D
I
don't
know
if
that's
going
to
be
acceptable
or
they're
going
to
say
that
on
the
cubelet
side,
you
you
get
the
query
and
then
you
change
what
gets
there,
but
you
know
again,
I
couldn't
decide
which
is
better
or
which
is
worse.
So
that's
why
I
just
said
I
want
to
get
this
reviewed
as
early
as
possible
so
that
we
have
time
if
I
need
to
get
it
fixed,
the
only
other
work
that
needs
to
be
done.
D
I
think
you
know,
in
addition
to
whatever
tim
and
and
jordan
asked
me
to
do,
is
our
end-to-end
test
and
I
think
fabian
has
already
started
working
on
it
and
he
had
mentioned
that
he
would
be
able
to
continue
to
help
out.
If
not
I
if
fabian
is
busy-
and
I
can
also
try
and
look
at
his
his
pr
and-
and
you
know,
move
that
to
using
the
new
query.
Endpoint.
A
I
would
hope
that
I'll,
like
a
lot,
I
I
think,
a
lot
of
the
feedback
that
especially
tim
was
giving
seemed
to
be
really
around
the
usability
of
the
the
ux
around
this.
So
I
would
hope
that
what
you
had
mentioned
like
how
how
it
actually
happened
under
the
hood
would
be
kind
of
some.
Some
some
leeway
could
be
there,
especially
since
this
is
going
to
be
an
alpha
feature,
because
I
don't
think
that
that
would
change
the
cube.
D
A
B
D
B
But
the
node
yeah,
the
node
doesn't
have
to
be
ready
or
you
know,
containers
could
be
crashing
and
that
type
of
thing
so
yeah,
that's
cool
and
then
can
you
did
is
is
etw
events
wired
in
at
all,
or
is
that.
D
Etw
events
are
not
wired
in,
but
it
could
be
wired
at
the
moment.
I
let
me
I
used
to
remember
this,
but
then
I've
I've
stopped
keeping
you.
D
Top
of
my
mind,
but
I
can
tell
you
that
in
windows,
the
current
thing
that
I'm
looking
at
is
the
application
logs.
So
I'm
doing
something
like.
A
Yeah,
because
with
the
etw
events
I
mean,
there's
there's
a
lot
of
different
aspects
of
having
uw
events
work
in
windows,
but
one
of
the
things
is
like
you
need
a
listener
to
listen
to
the
etw
events
and
and
then
serve
them.
However,
you
want
most
of
the
time
like
those
application
logs
are
the
application
is
emitting.
A
It
has
set
up
a
log
listener
and
is
emitting
etw
events.
I
don't
know
if,
like
I
I
I
don't
know
if
this-
if
it
would
be
appropriate
for
this
kind
of
functionality,
to
listen
to
those
etw
events.
But
if
there
are
things
already
like
log
collectors
for
etw
events
already
set
up,
then
we
could
scrape.
D
Sorry,
sorry,
I
I
think
we
can
definitely
augment
this
right,
because
the
the
same
way
we
are
going,
the
main
thing
is
like,
which
is
what
we
would
I
was
getting
pushback
from
tim
is
he
doesn't
want.
He
wants
a
query
to
be
really
simple.
So
underneath
the
hood
inside
like
where
we
are
creating
the
logging
command,
we
can
figure
out.
Okay,
do
I
go
to
the
application
logs
or
is
there
some
other
command
that
I
need
to
run?
I
think
that's
that
I
think
should
be
doable
yeah.
D
I
don't
think
that's
going
to
be
that
complicated,
because
already
now
I
I
try
and
figure
out.
I
run
a
command
to
figure
out.
Okay,
is
this
an
actual
application
on
on
the
windows
side?
Or
is
it
just?
You
know
a
bunch
of
log
files
and
if
that
command
returns
not
found,
I
basically
say:
okay,
let
me
go
and
try
the
files,
so
we
could
do
something
similar
if
we
want
to
go
to
other
logging
sources
to
figure
out
where
this
service
is
is
posting
their
logs.
A
F
I
j
I
joined
late
as
usual,
so
I
don't
know
if
I
I
don't
have
anything,
but
I
am
happy
to
look
at
the
the
windows
proxy
stuff
with
y'all.
F
But
yeah
thanks
arvin
for
all
that
log
stuff.
This
has
been
quite
the
odyssey.
F
I
have
one
thing
about
this
whole
thing
because
we're
adding
this
log
viewer
this
this
this
log
thing
and
then
this
agent
thing
keeps
coming
up,
and
I
have
a
general
question
about
windows
and
agents
and
because
there's
this
cluster
api
node
agent,
that
people
were
talking
about
at
one
point
and
then
openshift
has
one
and
we
brought
it
up
briefly
last
time-
and
I
guess
arvin
since
you're
here
has:
is
there
any
effort
anyone's
making
and
mark?
F
A
Are
you
talking
about,
or
I
I'd
say
signo
does
not.
I
think
cluster
api
is
probably
where
that
ownership
is
going
to
come
from,
and
I
think
james
has
a
little
probably
a
little
bit
more
up-to-date
information
than
I
do,
but
go
ahead.
Arvin.
D
Yeah,
all
I
know
is
there
are
things
inside
the
cubelet
that
is
cloud
provider
specific
and
that
is
being
moved
out
into
separate
agents
that
run
on
the
node.
I
think
it's
called
ccm.
A
F
A
James
there's
this
I
like,
I,
I
think
I
feel
like
this-
has
been
discussed
in
cappy
quite
a
bit
or
the
desire
for
one
right,
but
they're
always
wanting
to
leave
it
up
to
whoever's.
Setting
up
the
nodes.
F
It's
the
general
problem
of
you
know
so,
our
openshift
friends.
They
don't
rely
on
kube
adm,
they
have
their
own
agent
thing
and
then
in
cappy
we
rely
on
cube
adm,
but
we
have
a
race
condition
with
container
d
and
the
rest
of
the
world
has
been
doing
things
with
puppet
and
salt
and
all
that
stuff.
Where
you
get
an
agent
for
free
and
I'm
like
it's
really
weird
that
in
kubernetes
we
have
this
agent
called
the
kublet
and
it
doesn't
do
everything
we
want
it
to
do.
F
F
A
F
A
D
Oh
so
sebastian
demote
the
demon
that
runs
yeah,
so
you
could
call
it
an
agent
yeah.
D
F
F
Yeah,
it's
not
like
moving
it
back
to
the
kubelet.
It's
just
that
I
I
feel
like
we
need
something
like
mgmt
or
puppet
or
salt,
or
something
for
I
just
I
just
wish
we
had
something
that
we
all
agreed
on,
that
someone
else
maintained
that
we
could
use
as
the
glue
between
the
cube
adm
and
the
kubelet.
Like
a
thing,
you
know
I
mean.
F
F
C
F
Yeah,
it
was
something
something
like
that,
but
I
feel
like
something
comes
up
every
week
right
where
I
kind
of
wonder
this.
C
A
I
think
that
some
of
these
agents
both
have
like
aspects
that
are
not
really
coupled
to
the
whatever
infrastructure
provider
like
whatever
cloud
you're
built
on
to,
but
then
I
think
in
order
to
be
very
useful,
you'd
want
some
of
that,
to
you,
know,
reach
back
and
to
the
cloud
service
providers
and
get
information
and
and
do
some
extra
hookup
like
you
know,
attach
extra
disks
to
it
and
things
like
that,
and
that's
where
I
think
everybody's
going
to
have
a
different
opinion
and
it's
going
to
be
kind
of
hard
to
maintain
yeah.
F
I
know
I
don't
think
this
problem
will
ever
get
solved,
but
I
I
just
yeah
and
I'm
not
I'm
not
proposing
a
solution
either.
I
just
kind
of
like
I
feel
like
there's
this
weird
space
where
we're
modularizing
stuff
and
there's
stuff
that's
been
built
before
that
can
solve
a
lot
of
these
problems
that
we
have,
but.
F
Yeah,
well,
I
I
kind
of
don't
think
the
immutable
approach
is
ideal
for
windows.
I
don't
think
the
immutable
vm
built
by
image
builder
is
ideal
for
windows.
I
think
the
windows,
the
best
approach
for
windows
to
me,
feels
like
something
similar
to
what's
being
done
in
openshift,
where
you
you
have
a
vm
and
then
you
reach
into
the
vm,
and
you
configure
it.
F
Like
you
have
a
windows
machine
and
you
go
in
there
and
you
set
it
up,
but
there
doesn't
seem
to
be
any.
You
know.
F
A
Yeah
well
like
I,
I
will
point
out
that,
especially
for
getting
the
see
like
all
of
the
ci
jobs
set
up
where
we
want
to
put
in
custom
versions
of
the
accumulated
custom
versions
of
container
d.
Things
like
that,
we
do
have,
at
least
in
the
the
cap,
z,
workflows,
that
we're
using
have
a
lot
of
different
cases
where
we
have
pre
or
post
cubanium
actions
to
mutate
the
immutable
image
builder
image
yeah
on
the
fly
too.
A
So
it's
immutable,
but
we're
I
mean
in
to
some
to
some
extent,
but
the
vhd
is
that
you
base
it
from
is
mutable,
but
once
it's
on
the
system,
we
do
change
it
too.
So
I
I
can.
F
See
so
that's
the
yeah,
that's
the
model!
You
go
well,
you
know.
In
the
early
days
of
kubernetes
we
had
salt
and
we
did
all
the
installation
through
salt
and
it
was
like
we
got
rid
of
it
eventually,
but
I
kind
of
missed
those
days.
You
know,
but
it's
yeah
and
cops
is
doing
the
agent
thing.
I
think
the
rancher
folks
are
doing
an
agent.
I
think
everybody's
got
their
own
little
agent
and
everybody's
got
their
own
secret
agent.
That's
basically
doing
kind
of
the
same
thing.
C
E
Talking
about
azure
and
aks,
you
know
I
saw
this
article
go
out
a
couple
of
days
back
on,
like
moving.
You
know,
office
to
windows
containers
in
aks.
I
thought
that
was
pretty
big.
So
a
couple
of
questions,
so
is
this?
The
move
from
service
fan
windows
containers
only
even
says
fabric
to
aks,
and
if
so,
is
it
like
all
of
office?
You
know
parts
of
office,
you
know
if
you
can
share
some
additional
insights
on
what
moved,
because
I
thought
this
is
a
great
proof
point
for
customers
trying
to
adopt.
You
know:
windows,
containers.
C
Yeah
so
there's
all
of
the
back
end
services
for
office,
365
teams,
skype
pretty
much
the
whole
whole
range
of
things,
yeah
they're
running
their
their
entire
server
systems
on
windows,
containers.
A
And
and
they're
like
this
blog
article
that
I
just
opened
they're
they're
transitioning
from
service
fabric
right.
I
thought
that
was
the
case
or.
C
I
believe
so
in
some
instances,
but
not
all
yeah,
they
have
a
lot
of
different
services.
So
sorry,
who
is
that
cosmic
officer,
office,
365
and
teams
and
skype.
A
E
Yeah
just
saying,
especially
on
like
scale
and
perf
test
right,
obviously
running
the
back
end
for
office.
Skype
is
no
mean
feat,
and
you
know
I'm
sure
you
know
the
scale
goes
to.
You
know
tens,
if
not
hundreds
of
nodes,
so
any
lessons
learned
best
practices
and
you
know
how
you're
able
to
scale
scale
how
much
you're
able
to
scale.
I
think
you
know
we
presented
some
perf
numbers.
You
know
last
week
I
think
andrew
presented
some.
You
know
windows
and
openshift,
both
numbers,
but
you
know
any
indication
you
could
you
know
guide.
E
You
know
us
and
share
some
best
practices
and
how
you
know
you're
able
to
like
you
know,
deploy
you
know
a
lot
of
containers,
a
lot
of
pods
to
scale
the
backend
for
office,
especially
skype.
You
know
the
traffic
can
be,
you
know
quite
slightly
up
and
down
so
any
best
practices
around.
That,
I
think,
would
greatly
help
the
community.
C
Yeah
absolutely,
and
we
have
a
lot
of
internal
services
that
are
also
just
moving
onto
windows
containers,
so
we're
planning
on
putting
out
a
whole
bunch
of
information
and
and
guides
on
how
to
do
these
things
in
the
future.
So
definitely
awesome.
A
And
I
know
one
of
the
one
of
the
windows:
qcon
talks
is
hosted
by
some
engineers
at
relativity
and
I
believe
that
the
talk
was
similar
to
that
how
how
they
handled
some
scale,
issues
like
just
scalishes
that
they're
into
and
how
they're
handling
managing
massive
workloads
and
windows
but
yeah.
Let
me
let
me
see
if
we
can
get
somebody
like
this.
This
mark
power,
other
people
to
come
and
give
some
guidance
and
then
sorry
james.
I
think
I
cut
you
off.
B
I
was
gonna
say
the
same
thing.
It
was
like
there's
that
talk
at
kubecon
and
then
there's
a
few
other
blog
posts
out
there
as
well,
that
have
been
rolling
out
in
the
last
couple
months
that
are
similar
and
talk
about.
You
know
other
success
stories
on
windows
containers
as
well,
so.
A
Let's
yeah:
let's
see
if
we
can
yeah
brandon,
let's
see
if
we
can
get
some
to
be
a
little
bit
more
like
I'm
just
going
through
this
one
right
now
and
this
one
seems
kind
of
high
level,
but
maybe
some
more
you
know
detail
focus
like
tips
might
be
helpful
too.
E
Yeah
yeah,
we
need
like
a
landing
page.
You
know,
maybe
in
the
upstream
repo,
with
all
like
talks,
you
know
presentations,
I've
been
maintaining
one,
for
you
know
the
local.
You
know
windows
on
openshift
and
you
know
glad
to
share.
You
know
a
lot
of
the
information,
it's
all
public
as
well,
but
we
probably
need
to
aggregate
some
of
these
things
from
you
know
multiple.
You
know
companies,
so
you
know
it's
a
good
landing
spot
for
windows
containers
and
I
think
there
was
some
talk
from
you
know
what
like
doing
like
a
containers.
E
A
Yeah
I've
been
maintaining
a
youtube
playlist
of
all
of
the
like
windows,
related
kubecon
or
just
other
conference
talks
too.
I
will,
I
think,
that
it's
on
the
meeting
notes,
but
I
close
that
tab
I'll
or
like
at
the
top
I'll
add
that
again
too,
but
yeah.
This
is.
These
are
all
great
ideas.
B
F
With
windows
containers-
and
maybe
we
can
get
one
of
them
to
talk
about
what
they're
doing.