►
From YouTube: Kubernetes SIG Windows 20220215
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
A
Okay,
in
terms
of
announcements,
I
guess
still,
the
most
important
upcoming
milestone
is
still
124
code
freeze,
the
which
is
march
30th,
2020
2..
I
guess
the
other
announcement
is.
I
believe
that
after
the
next
set
of
patches
patch
releases,
120
is
going
out
of
service
is
so
if
anybody's
still
running,
120,
consider
updating
clusters
and
I
think.
B
A
And
I
think
with
that
was
that
the
last
release
that
we
were
supporting,
one
of
the
sac
releases-
do
you
remember
james
is
one
of
the
so
does
that
roll
off
one
of
the
sack
releases
that
we
were
supporting
I'll
have
to
check.
We
can
check
that
next
week.
I
think
at
this
point
we're
supporting
20h2
only
yeah
and
when
I
served
2019
and
2022
still
that's
yeah.
If
anybody
has
any
announcements,
feel
free
to
add
them
or
call
something
out.
Another
announcement
is
prometheus.
The
windows
exporter
support
has
merged.
A
This
was
kind
of
a
long
process,
but
it's
another
kind
of.
I
think
success
in
bringing
feature
parity
between
windows
and
linux
up
since,
and
I
think
a
lot
of
people
have
been
asking
for
this
and
using
this,
and
james
also
was
going
to
do
a
demo
today.
So
I'll
save
some
of
that
for
the
demo.
A
Does
anybody
have
anything
else
to
talk
about
for
the
agenda?
If
not,
I
can
hand
it
over
to
james
and
you
can
demo
some
of
that
work.
C
So
thanks
for
approving
that
and
getting
that
merged
in
so
that'll
allow
you
to
run
as
a
non-root
user,
because,
right
now
you
have
to
bind
to
443,
which
can
only
be
done
as
root,
and
even
if
you
try
to
do
capabilities,
they
didn't
work
to
do
that
bonding.
So
now
you
can
at
least
configure
that
if
you
need
to
run
it
as
a
non-root
user
now
how
well
that
works
or
if
there's
any
edge
cases
around
that
I'll
find
out
basic
stuff
seem
to
work
just
fine.
A
And
I
think
there
is
another
there's
a
pr
to
deploy
all
of
this,
but
the
helm
chart
too.
So
if
anybody
is
interested
feel
free
to
take
a
look,
james
and
myself
have
been
slowly
reviewing
it.
I
think
that'll
help
ease
the
deployment
instead
of
just
asking
people
to
finally
run
a
script
that
doesn't
want
things.
C
Yeah
and
that's,
I
did
some
more
work
with
just
getting
to
make
stuff
set
up
yesterday
and
push
that
up,
and
then
you
know,
there's
a
couple
other
things.
So
if
there's
some
configuration
that
you
don't
agree
with,
if
you
look
at
it
just
let
me
know,
and
we
can
make
some
tweaks,
I
gotta
get
the
testing
stuff
set
up
and
some
of
the
automation
and
that's
left
on
that
one.
D
So
I
just
have
one
question
mark:
this
is
regarding
the
content
id
cna
support.
I
think
I
started
a
thread
where
james
is
helping
and
we
are
trying
to
discuss
if
we
are
missing
any
comments
in
the
window.
Overlay,
I
think,
looks
like
sdn
overlay
works,
so
we
are
trying
to
discuss
that
with
james.
I
think
david,
I
guess
tomorrow.
D
Probably
we
will
keep
you
guys
posted
on
the
progress
after
the
discussion.
So
maybe
something
is
missing,
whatever
the
changes
mentioned
in
the
sdn
overlay
and
we
may
need
to
make
put
the
changes
or
make
similar
changes
in
window
overlay.
James,
please
correct
me.
If
I
misunderstood.
A
D
I
I'm
not
sure
so.
I
was
just
trying
to
chat
with
the
channel,
especially
we
would
like
to
talk
to
james
and
do
it.
So
I
don't
know
if
anybody
interested
I'm
open
to
send
the
invite
as
well.
D
D
E
It's
at
the
moment
on
on
google
me,
but
we
can
move
it
to
zoo
zombie
wherever
it
is,
it's
fine.
I
don't
think
we.
D
What
I
can
do
is
I
can
just
post
that
in
a
new
thread
or
the
same,
our
meeting
in
white
information.
So
if
anybody
else
wanted
to
attend.
A
D
Also,
our
just
another
question:
sorry,
I
don't
mention
that
another
person
from
microsoft
was
heavily
involved
in
this
doing
overlay.
So
not
sure
is
it
possible
for
him
to
attend
the
meeting,
I'm
not
sure
he's
in
thread.
He
tagged
him
ivan
tagged
him,
but
just
want
to
make
sure
that
if
he
get
the
notification
or
he's
not
reachable,
do
you
guys
have
any
idea
who
is.
E
That
mark
mark
ellen
was
talking
about
jocelyn.
E
Yeah,
the
reason
is:
there's
this
hybrid
overlay
component
that
we're
using,
and
that
was
heavily
worked
on
by
joslin.
F
Yeah,
I
see
okay
yeah.
I
think
the
person
that
that
I
added
keith
would
be
responsible
for
continuing
that
work.
F
Yeah,
unless
he's
already
added-
I
I
know
I
added
him
to
the
original
email
fred.
I
thought
everyone
also
yeah
keith
should
be
there
definitely
yeah.
D
G
D
I
remember
you
added
some
container
knit
that
microsoft
that
that
was
bouncing
back
for
me.
It
says
like
it's
not
reachable,
so
let
me
check
it
also
and
at
the
meantime,
I'm
going
to
post
that
meeting
in
the
slack
channel
of
syd
windows
channel.
So
if
anybody
else
want
to
talk-
and
they
can
attend
this
one
I'll
do
that.
A
A
H
One
is
you
know
last,
I
think.
A
couple
of
weeks
ago
we
discussed
around
the
idea
of
a
windows
unconference
just
trying
to
convey
to
the
community
that
red
hat
is
very
interested
in
you
know
the
agenda
and
you
know
helping
build
out
and
supporting
whatever
way
and
so
glad
to
you
know,
work
with
whoever
is
going
to
be.
You
know,
driving
that.
A
Yeah
so
waitron
she
gave
this
from
our
end
was
the
one
who's
I
think
going
to
be
driving
that,
and
she
is
continuing
to
do
that.
I
think
she's
been
asking
some
of
the
folks
who
have
posted
previous
unconferences
like
that,
before
kind
of
where
what
where
to
get
started,
but
we'll
continue
to
kind
of
post
updates,
as
as
they
come,
I'm
glad
to
hear
that
there's
more
people
willing
to
or
that
there's
interest
in
this
area.
H
H
20
22.,
I
believe
you
know
you
were
going
to
update
the
upstream
docs
with
some
known
bugs
known
limitations.
You
know
just
want
to
make
sure
you
know
that's
going
to
be
done.
A
I
I
just
want
to
confirm,
because
I
think
windows
defender
sometimes
has
this
performance
issue
and
I
think
that's
why
here
they
disabled
some
part
of
it.
I
just
want
to
confirm
from
like
a
windows
expert.
I
This
is
recommended
or
any
security
concern
or.
A
Let
me
take
a
look.
I
think
that
we've
recommended
to
disable
or
to
like
add
exclusions
for
like
well-known
and
especially
signed
binaries.
I
have
to
follow
up
on
this
buzz.
I'm
probably
gonna.
Send
this
your
way
offline.
J
A
Okay
sounds
good
all
right
j,
we'll
get
back
to
you
on
this.
K
Okay,
so
I
guess
so:
there's
prometheus
windows
exporter,
it's
meant
to
run
on
the
windows
node
and
it
collects
a
bunch
of
different
metrics
about
the
node,
including
containers
and
cpu
and
all
sorts
of
stuff,
and
it's
configurable
to
collect
various
stats,
probably
about
a
year
and
a
half
ago,
maybe
two
years
ago
somebody
opened
up
a
pr
to
add
support
to
this
via
to
inside
kubernetes
the
the
issue
that
that
was
happening.
K
There
was
that
the
when
you
run
the
windows
exporter
inside
a
container
doesn't
actually
have
access
to
the
host
because
of
process,
isolated
containers,
and
so
now
that
we
have
host
process
containers,
we
can.
We
can
run
the
windows
exporter
directly
on
the
host
and
get
support.
So
I
picked
up
the
pr
and
kind
of
brought
it
along
and
gave
it
post
process
support
here,
and
so
this
is
the
document
on
how
this
is
actually
done.
There's
a
few
notes
that
you
need
to
have
kubernetes
122.
K
You
have
to
have
a
container
d
1.6
beta
version,
and
then
you
also
need
to
have
the
windows
host
process
container
feature.
Gate
turned
on
for
all
this
to
work.
The
repo
has
the
ability
to
build
this
container.
So
if
you
want
to
build
your
own,
you
can
use
this
make
command.
The
one
thing
I
noticed
was
this
merged
the
end
of
last
week.
I
think,
and
the
the
prometheus
windows
exporter
ci
didn't
actually
build
the
like
nightly
build
of
this
container.
So
it's
not
available.
K
So
I'm
using
the
one
that
I've
built
myself.
I've
opened
up
an
issue
to
get
that
resolved
with
the
windows.
Exporter
team.
They
just
switched
a
bunch
of
ci
around,
and
so
it's
all
fairly
new,
but
without
further
ado,
we'll
just
kind
of
jump
in
so
the
windows
exporter
is
now
a
daemon
set.
So
it
runs
on
all
of
the
windows
nodes
and
it's
configured
so
that
it
only
runs
on
the
windows.
K
It.
The
the
container
itself
mounts
the
config,
so
you
can
pass
various
config
into
the
windows
exporter
and
enable
different
types
of
collectors
and
different
services
that
you
want
to
run
against,
and
so
that
config
is
a
config
map
in
kubernetes
and
it
gets
loaded
as
an
arg
for
the
host
process
container.
K
And
then
this
is
the
image
it
will
be.
An
official
image
once
we
get
that
build
process
resolved
the
you
can
see
here
on
the
spec.
We
have
the
security
context,
windows
host
option
host
process
and
then
we're
just
running
it
as
system
so
that
we
have
full
access
to
all
the
various
metrics
that
we
need-
and
this
is
this
isn't
strictly
necessary,
but
it
helps
make
sure
the
firewall
rule
is
opened
up
on
the
on
the
host
for
this
particular
port.
You
can
drop
this.
K
If
you
do
that
prior
to
that,
and
then
once
that's
applied,
you
can.
What
I'm
using
is
a
tool
called
cube,
prometheus,
which
is
the
prometheus
operator
set
of
scripts
and
dashboards
and
config
configuration
here,
and
so
I've
installed
this
on
the
machine
and
I'll
show
this
to
you
in
a
second
once
that's
configured,
you
can
create
one
of
these
windows.
K
Exporter,
pod
monitors-
and
this
just
says
this
is
the
name
of
the
job
and
then
it
matches
any
pods
that
have
this
particular
label
and
it
scrapes
the
endpoint
for
that
pod,
and
so
now
I'll
go
over
to
my
cluster.
I
have.
K
Here
you'll
see
oops,
oh
it's
because
I
zoomed
in
let's
do
that
again.
Okay,
so
you'll
see
in
the
monitoring
space
I
have
the
prometheus
running
I
have
graphon
is
running
the
node
exporter
on
on
linux
is
running
all
of
that
stuff.
That
comes
out
of
the
prometheus
operator
and
the
cube
prometheus
project
is
all
already
running,
and
so
I
can
now
apply
the
windows
exporter
daemon
set,
and
if
I
do
that
now
we'll
see
there's
two
pods
that
get
created
here.
K
The
that's
because
I
have
two
windows
nodes
up
here
that
are
that
are
windows,
configuration
that
are
windows
nodes,
and
so
it
ran
one
on
each
on
node.
And
if
I
do
that
now,
they're
now
both
running
so
I
can
do.
K
Switch
into
the
monitoring
namespace
and
do
k
logs
windows,
exporter
and
here
you'll
see
that
the
exporter's
actually
running.
This
is
it's
loading
that
configuration
file
and
for
anybody
that
hasn't
run
host
process
containers.
They
get.
K
This
special
draw
mount
folder
in
the
host
and
that's
what
this
kind
of
crazy
long
guide
is
and
then
one
of
the
things
that
we
added
to
the
prometheus
was
just
to
say:
hey
are
you
running
as
who
you're
actually
running
as
because
it
would
be
easy
to
build
this
container
and
not
apply
the
host
process
configuration
and
it
will
give
you
a
warning
if
you're
running
as
container
administrator
or
container
user,
because
if
that's
likely
that
you
are
running
inside
a
process,
isis
isolated
container
and
not
as
a
host
process
container.
K
L
K
K
I'm
sure
I'm
poor
ordering
the
right
thing
there
we
go
so
just
took
a
second
to
find
it.
It's
an
unknown
state
and
there'll
be
a
scrape
here
in
a
minute
and
we'll
start
to
collect
the
windows
metrics
so
yep.
So
there
we
go.
It
found
the
two
nodes
and
it's
now
scraping
those
from
those.
So
we
see
that
the
the
instances
1.06
so
that's
my
node,
the
the
ip
addresses
of
my
nodes.
K
So
if
I
do
k
get
modes
wide,
I
should
see
this
is
the
windows
node
and
I've
got
104.
This
is
my
other
windows
node
and
it's
106..
So
if
I
now
come
back
here
I
can
see
106
and
104.
K
K
I
can
this
windows
cpu.
Total
time
is
one
of
the
the
metrics
that
the
node
exporter
puts
out,
and
so
I've
pinned
it
to
the
106,
and
you
can
see
I'm
starting
to
pick
it
up
over
here
and
this.
This
is
practice
for
me
doing
earlier,
make
sure
that
this
this
worked,
and
so,
if
we
there's
another
metric
that
I
can
there's
a
bunch
of
different
metrics,
but
if
we
do
windows,
cpu
usage
total
seconds,
this
is
another
one
that
gets
put
out
by
the
windows.
K
Exporter
and
you'll
see
there's
two
containers
here.
These
two
containers
are
the
pause
container
and
an
is
application
that
I
was
running.
You
probably
can't
see
it
here,
but
the
container
id
is
6b
5,
9
bd,
and
if
I
and
this
one
up
here
was
db49,
you
probably
can't
see
that
it's
pretty
small.
But
if
I
come
over
to
the
node
that
these
are
running
on,
I
can
do
hcs
diag
list
and
I'll
see
both
of
those.
K
K
Is
this
one
right
here,
so
the
six
paws,
so
those
are
correlating
to
the
metrics
that
are
being
collected
for
those
containers,
so
the.
K
So
all
those
metrics
that
you
that
this
thing
collects
are
all
listed
on
the
front
thing
here,
so
you
can
see
them
all.
The
containers
here
is
in
here
and
then,
if
you
use
the
prometheus
mix-it
mix-in
there's
a
bunch
of
rules,
you
can
actually
get
like
hey
it'll
tie
all
the
pods
together
all
the
containers
into
pods.
So
you
can
actually
get
nice
pretty
graphs
on,
like
the
metrics
of
your
pods
and
everything
so.
K
It
adds
a
bunch
of
prometheus
rules
that
map
those
that
add
labels
to
the
container,
so
it'll
tie
a
container
to
a
pod
and
so.
B
K
Yep,
so
to
add,
like
you
can
see,
I
think
there's
a
pod
in
here
yeah.
These
are
the
pod
rules,
so
you
could
say
windows
pod
container
available
and
you
would
actually
see
the
number
of
pods
that
are
running
inside
your
on
on
each
node
and
you
can
make
a
nice
pretty
graph
with
that.
This
actually
does
that
all
out
of
the
box
for
you,
if
you
apply
these.
K
So
if
you
go,
I
didn't
deploy
that
mix
in
here,
but
if
you
go
into
docs
in
windows
here
this
is
this:
tells
you
how
to
do
that
and
then
there's
also
a
full
example
that
applies
those
metric
rules
to
it.
There
isn't
one
of
these
for
host
process
yet
so,
if
anybody's
looking
for
improvement,
I
did
open
an
issue
to
improve
this
for
this
particular
project,
but
I
didn't
get
around
it
this.
This
uses
the
old
way
of
installing
it
directly
on
the
host
and
using
static
configs.
K
Whereas
now
that
we
have
the
metric
server
actually
running
in
the
pod,
we
can
scrape
it
and
dynamically
build
that
configuration.
E
B
E
That
those
those
rules
are
really
important,
because
I
think
a
lot
of
people
will
run
windows
exporter
and
then
expect
everything
to
work
magically
the
way
it
works
for
linux.
Sometimes
you
have
to
map
certain
things
that
are
windows
specific,
to
make
it
a
little
bit
more
generic,
so
that
when
you
go
and
look
at
your,
I
don't
know
whatever
dashboard
that
you're
using
it.
It
actually
matches
up
with.
What's
there
in
in
linux.
K
Yeah,
once
you
install
those
things
you
get
this
pretty
graph
like
this,
like
cpu
and
memory
and
all
sorts
of
cool
stuff,
and
already
somebody
somebody
from
your
team
also
helped
out
with
some
of
this
cute
prometheus
stuff
as
well.
I
think.
E
Yeah
that
was
yeah.
That's
how
we
found
out
that
oh,
we
have
to
create
these
rules
to
kind
of
sort
of
normalize.
All
of
this,
and
some
people
forget
to
do
this
right.
They
they
install
windows,
exporter
and
and
they
don't
apply
the
rules
and
then
things
don't
appear
the
way
it's
supposed
to
yeah
all.
K
B
Is
there
a
like
a
thing
that
ties
all
this
information
together,
james,
like
a
dot
diagram
or
something
that
sort
of
like
because,
like
you
know,
every
time
I
look
at
prometheus,
it
gets
more
and
more
complicated
and
I
used
to
understand
it.
But
now
I
just
don't.
I
can't
even
remember
all
these
bits
anymore
like
to
to
just
get
a
basic
prometheus
working.
You
have
to
you're
accessing
api
server.
Metrics
endpoint,
like
you
know,
I
mean,
and
then
you're
like
you're
just
there's
so
many
little
weird
things
that
get
connected.
A
K
Yeah,
I
think
so,
if
you
want
to
use
the
prometheus
operator,
if
you
wanted
to
install
it
through
a
helm,
chart
or
something,
I
don't
know
exactly
how
to
do
that.
I
haven't
experimented.
B
K
B
Okay,
cool-
I
I
just
was
yeah,
I'm
just
wondering
yeah
ever
since
they
came
out
with
the
prometheus
operator,
I
think
things
got
a
lot
more
confusing
because
they
stu
people
stop
documenting
the
manual
way
to
do
it.
But
maybe
the
right
thing
for
me
to
do
is
just
to
grow
up
and
start
using
this
operator.
What
do
you
use
arvind?
Do
you
use
the
operator
over
there
at
red
hat.
E
E
So
that's
all
part
of
day
one
installed
the
cluster,
so
we
don't
have
to
do
anything
as
a
as
a
manual
step.
It's
it's
because
it's
these
are
collecting.
What
do
I
say:
node
metrics
they're
needed
when
you
install
the
cluster
on
day,
one
so
they're.
It's
already
present
and.
E
E
Details
about
it,
but
what
we
have
to
do
for
windows
is:
we
have
to
apply
those
similar
rules
that
that
james
was
showing
and
that
I
think,
I'm
trying
to
remember
it's
been
a
while,
since
we
we
did
this
either
we
applied
as
part
of
our
operator,
or
I
think
it's
done
as
as
part
of
the
the
main
operator
that
is
being
installed.