►
From YouTube: Kubernetes SIG Node 20220607
Description
SIG Node weekly meeting. Agenda and notes: https://docs.google.com/document/d/1Ne57gvidMEWXR70OxxnRkYquAoMpt56o75oZtg-OeBg/edit#heading=h.adoto8roitwq
GMT20220607-170455_Recording_1920x1032
A
Hello:
everybody,
it's
June,
7th
2022,
it's
a
signals.
Weekly
meeting
welcome
everybody
today
is
the
biggest
item
on
agenda
is
to
go
through
the
caps
for
125
and
let
that
donut
Derek
fill
up
the
Caps.
We
want
to
track
for
125
and
release
trekken
thingy
spreadsheet,
so
Raven
thanks
for
sharing
the
screen
and
let's
get
going
and
go
see
what
caps
you
want
to
deliver
in
195.
A
B
Myself,
I
I
think
I'm
a
kid
now.
Let
me
just
to
reshare
perfect.
B
B
A
Give
me
some
time:
yeah
I,
think
the.
D
B
Cool
thanks
yeah
so
anyway,
sorry
about
that
I
just
put
a
finalist
cap
table
on
top
of
this
Dock,
and
this
is
basically
a
sum
of
the
planned
caps
listed
below
minus
the
ones
that
do
not
have
clear
owners
or
the
ones
where
people
say
they
don't
have
capacity
to
work
on
so
I
just
compiled
this
finalized
planning,
caps,
I
guess
for
now
we
can
just
go
through
the
list
and
to
see
what
is
well.
B
Is
there
any
change
we
should
make
on
this
list
and
if
not,
we'll,
just
put
this
list
into
our
that
finalized
release
spreadsheet
so
yeah
I?
Guess,
let's
go
through
them.
The
first
one
in
place,
updates
of
pod
resources,
vinay
I
think
this
is
being
actively
worked
out
right.
We
should
keep
it
here.
Definitely
yeah!
It's
been
in
here.
E
Yep,
that's
an
update
for
it
in
the
meeting
today,
but
vinay
is
out.
B
C
C
A
Next
one
I
want
to
try
if
it
will
be
approved
by
sick
testing.
It
will
be
in
125.
B
Okay,
so
this
is
like
conditionally
in
125,
right,
yeah.
B
All
right
support,
username
space.
B
Cool
frenzy
container
checkpointing.
B
C
B
Dynamic
resource
allocation,
I
think
we
we
don't
have
a
sorry.
We
don't
have
a
status
here,
but
we
have
some
owners
here
listed.
C
C
G
B
We'll
do
that
moving
on
part
conditions
around
starting
completion
of
sandbox
creation.
This
is.
C
B
Yeah,
the
next
one
device,
plugin
API
feature
so
in
says
it
needs
to
for
we
need
to
follow
if
it
is
being
accurately
worked
on.
So
so.
C
This
is
so
zwanko
reached
out
to
me.
He
wants
to
work
on
it
and
I
think
he
he's
found
reviews
approvals.
I'll
ask
him
to
update.
H
Yeah,
it's
a
yeah,
this
particles
talking
the
owner
so
yeah.
It's
well
a
discussion
topic
again.
So
in
terms
of
like
code,
it's
it
could
be
really
small
if
you
just
get
to
agree
that
let's
go
with
this
and
get
good
reviews
about
them
so,
but
I'm
actually
working
on
that.
This.
A
To
agree
on
scenarios
first-
and
it
was
some
like
it
was
a
request
I
believe
maybe
my
memory
doesn't
serve
me
right,
but
I
believe
there
was
a
request
to
list
scenarios
where
it
will
be
used
and
what
other
like
generalization
we
can
apply
there.
H
But
then
there
was
at
least
I
remember,
request
to
kind
of
show
about
show
how
the
kind
of
existing
Book
value.
A
Even
if
I
have
one
week
left
before
the
enhancement
fees
will
we
be
able
to
find
the
reviewers
and
approvers
I.
C
C
C
That's
what
I
feel
because
like
torn
and
Derek
are
not
around
today
and
we
may
be
running
out
of
time
to
just
make
a
decision
here.
A
Okay,
let's
remove
from
125
and
okay,
you
can
always
keep
working
on
that
it.
Just
we
wouldn't
work
on
it.
Just
had
this
shouldn't
be
shipped.
G
So,
okay,
if
possible,
let's
keep
it
for
one
week
and
if
we
will
find
clear,
viewer
and
I'll
approved
forward
during
this
week,
when
we
can
oh
later.
D
A
B
G
But
I
think
it
will
require
just
some
ad
hoc
conversation
with
the
people.
H
A
Hey
so
this
is
being
still
being
discussed
with
the
Sig
storage
I
think
we
can
remove
it
from
Sig
node
for
125.
J
E
It's
on
my
list
for
tomorrow,
so
it
looks
like
it
got
merged,
though
so
that
should
be
good.
Okay,
okay,.
B
Yeah,
the
Caps
first
right
support,
username
spaces.
I
think
this
one
is
under
review
already,
so
this
should
be
good.
B
B
B
For
Windows
privileged
containers,
we
have
owner
reviewers,
so
this
one
should
be
good
right.
F
A
Okay,
I
think
biggest
concern.
There
was
misunderstanding:
like
I'm
misaligned,
this
GA
will
change
something
to
be
written
in
cap,
so
there
is
a
text.
Change,
Mark,
I
think
this
needs
to
be
debuted
in
the
pr.
J
B
Okay,
number
is
still
shut
down,
I
saw
Xin,
Yang
created
a
cap
for
this
one,
so
I
just
assumed
she
would
be
the
owner.
C
Try
to
Ping
her
so
we'll
have
to
check
if,
if
the
cap
is,
this
is
an
update
to
the
kit
for
beta.
B
Yeah
she
added
that
a
milestone
125,
so
I
just
assumed
that
yeah,
yes,
yeah.
B
A
B
B
Yeah
about
that
one
I
might
have
some
capacity
to
take
it
over.
It's
that's
fine,
I
I
know
that
Andrew
is
out
so
I.
Think
the
next
step
for
this
one
is
to
promote
it
to
GA
right.
E
Yeah,
not
I,
don't
quite
remember,
What's
blocking
it.
Now
it's
no
bit
in
my
head
for
a
few
weeks,
but
a
DT
is
out
for
the
next
few
months
and
Matthew
doesn't
have
time.
So
if
you
want
to
take
it
over,
that
should
be
fine.
Okay,.
B
All
right
pod
priority
based
on
another
group,
will
shut
down.
C
B
Okay
demonstrated
support
Master
to
improve
workload,
availability
yeah.
So
this
one
is
I
right.
I
came
here
because
it
says
no
shoot
in
what
signal
should
support
this
work
and
yeah,
but
we
don't
have
a
clear
owner
on
Signal
site.
A
B
Yeah,
it
says
controller
API
should
drive
so
yeah
I,
don't
know
if
we
should
even
keep
it
here.
E
B
B
Okay,
ephemeral
containers
have
owner
where
we
were
should.
A
It's
good
to
go
I
just
what
is
it
if
you
don't
have
a
ga
requirement
and
criteria,
then
how
we
decide
whether
it's
raised
to
G,
who
is
the
owner.
J
A
Maybe
we
need
to
bring
this
person
and
see
if
this
actually
will
move
forward.
B
B
Status
on
this,
okay,
moving
on
enables
sitcom
by
default,
have
owner
review
approvers.
So
this
one.
B
It's
speed
up,
recursive,
see
Linux
label
change
similar.
This
one
is
on
mostly
on
storage
and
we
have
a
signaled
contact.
I
guess
we
don't
need
to
keep
this
here
as
well
right.
B
Yeah,
just
in
general,
yeah,
I
I,
don't
think
we
need
to
keep
like
a
reviewer
review
work
in
this
in
this
track
list.
Correct
me
if
I'm
wrong,
that's.
E
Helpful,
that's
a
reminder,
given
that
we're
also
kicking
off
a
bunch
of
work
on
testing
and
maintainability,
just
to
be
cognizant
of
how
much
we're
signing
up
for.
C
B
B
A
Is
it
on
track
for
125,
because
I
don't
think
it
was
really
good.
B
C
D
B
Right,
we
have
owners
and
oh,
we
don't
have
reviewers
approvers.
Yet.
C
D
B
Okay,
moving
on
yeah.
A
I
had
this
cap
here
and
somebody
proposed
it.
It's
very
small.
C
A
Okay,
so
just
maybe
we
need
to
have
another
column
marking
something
that
we
clearly
know
that
it's
125
so
don't
know
that
I
can
move
to
release
track
and
spreadsheet
and
we
will
have
the
rest
of
caps
reviewed
next
week,
just
before
enhancement,
release,
yep.
B
Sounds
good
I'll
do
that
I
can
well
add
a
new
column
to
the
table.
A
Okay,
thank
you
Raymond
for
driving
this.
We
have
yeah
I
also
proposed
to
have
soft
fees,
as
we
had
two
previous
releases
already
I
suggest
July
12th
as
a
soft
release.
It
will
give
roughly
three
weeks
after
enhancement
fee,
so
like
almost
five
weeks
from
now
for
beta
in
graduations
and
GA
implementation.
So
last
time
we
discussed
that
software
freeze
will
be
different
for
beta
and
Alpha.
If
caps
are
Betina
and
bet
in
graduation
should
be
fully
merged
by
a
soft
threes.
A
If
they're
not,
then
we
need
to
remove
them
from
from
a
release,
because,
if
they're
not
immersed
yet-
and
we
don't
have
any
like
extra
time
to
have
them
soaked
in
a
RCI
environment
and
have
it
tried
out
if
you
merge,
big,
GA
or
beta
feature
just
before
release
it,
the
quality
may
not
be
ideal
and
for
all
the
features
that
are
not
enabled
by
default.
A
We
typically
allow
more
time
so
soft
threes
is
just
for
PR
to
be
out,
so
we
know
that
somebody's
working
on
that,
but
if
PR
is
not
merged.
Yet
it's
also
fine,
because
it's
Alpha
that
is
not
enabled
by
default
and
quality.
May
be
a
little
bit
lower,
so
we
find
this
delaying
it
to
the
code
freeze
date,
if
anybody
have
any
problem
with
this
date
or
suggestions
how
to
improve
please,
but
you
know-
and
otherwise
we
will
announce
this
date
and
make
it
visible
to
everybody.
A
Okay,
yeah
this
cap,
I
just
put
in
the
tracking
sheet
and
now
Marcus.
H
Yeah
once
again
about
this
class
resources,
so
probably
I.
I
I
A
A
H
What
I'm
talking
about
this
class
resources,
which
is
a
bit
unfortunate
name,
carries
a
lot
of
like
Legacy
from
the
first
steps,
but
anyway,
let's
keep
this
name
for
now,
so
first,
to
remind
what
what
this
class
resource
is
kind
of
would
be
able
would
be
about,
or
about
so
kind
of
keywords,
resources
that
require
no
allocation
or
result.
H
Reservation
capabilities
in
kubernetes,
a
kind
of
non-accountable,
infinite
resources,
and
one
kind
of
important
property
in
this
is
that
multiple
containers
can
can
share
kind
of
one
one
resource
or
class,
so
think
about
think
about,
for
example,
how
that
all
keywords,
classes
works
or
any
any
containers
can
be
in
the
same
Port
keywords
class,
then
why?
Why
would
we
like
to
add
support
for
this
is
that
they
would
allow
to
support
some
resources
that
are
kind
of
inherently,
for
example,
in
Hardware
class,
based
with
a
limited
number
of
classes.
H
So
that's
really
no
limited
number
of
classes
so
that
not
every
container
can
can
have
its
own
class
and
other
cases
I'm.
One
of
one
example
of
this
is
this:
Intel
rdt
technology
for
caution,
memory,
bandwidth
allocation,
so
some
hardware,
for
example,
you
may
have
limit
for
only
eight
classes
available.
So
that's
not.
It
is
possible
to
have
one
class
per
container
thing
very
practical
scenario.
H
H
So
an
example
is
this
block
I
o,
where
we
implement
this
class
based
approach
for
local
or
ioc
group
controller
configuration
so
resources
that
are
kind
of
further
configuration
parameters
are
very
node
dependent
than
kind
of
cluster
admin.
I
know
how
to
set
up
some
set
up
the
parameters,
but
users,
especially
in
heterogeneous
systems,
have
cannot
practically
practically
give
any.
H
Meaningful
parameterization
because
of
the
Port
landing
on
any
node.
H
And
the
kind
of
design
principles
in
this
kept
we're
talking
about
is
that
the
class
resources
would
be
opaque
as
possible
to
kubernetes
so
kind
of
opaque,
non-inter
integer
resources
to
kind
of
draw
a
line
between
these
and
extended
resources.
Currently,
so
configuration
management
would
be
handled
by
the
container
runtime
on
the
underlying
layer
and
kubernetes
good
would
not
care
about
these
details
virtually
at
all
and
the
CRI
app
API
changes
proposed
here
are
very,
very
minimal.
Just
and
then
you
start
New
Field
in
in
resource
container
resources.
H
Then,
the
last
time
we
discussed
there
was
this
kind
of
question:
how
how,
for
example,
how
the
block
I
o
support
would
look
like
from
the
user
point
of
view,
so
I'll
have
a
really
really
short
demo
about
this
one.
So
foreign,
do
you
see
anything
probably
too
small.
H
Okay,
so,
basically
now
in
this
demo,
I
have
a
container
they
attached
to
support
this
class
resources
like
the
CRA
API
and
kubernetes
125,
three
125
as
well
support
this
and.
H
H
I
H
Cluster
running,
in
the
background,
so
just
to
show
that
we
have
one
node.
Oh,
let's
actually.
H
H
H
Then
how
to
use
this
with
the
annotations.
It's
a
simple
example,
with
with
three
content
called
with
three
containers
container
one
two
and
three
with
annotations:
okay,
I
forgotten
that
I
put
also
one
rdt
cluster.
That's
not
content
on
that
one,
but
but
with
this
first
annotation
we
set
the
kind
of
Point
pod
level
default
at
all,
containers
go
to
normal
class
and
then-
and
the
second
example
would
say
that
container
one
goes
to
hyper
analytics
class
of
this
block
resource
and
then,
if.
H
But
now
the
what
has
been
deployed
and
then
the
kind
of
parameters
that
we
set
are
are
now
visible
in
in
csfs.
So.
D
H
According
to
our
configuration.
I
H
No
not
at
this
point
I
mean
it's
it's.
There
are
a
lot
of
future
steps
like
the
resource
list,
cover
and
resource
tattoos
and
scheduling
and
resource
quota
and
stuff
like
that,
but
they
are
in
the
future
steps
of
the
tip
so
I'm,
basically
a
proof
of
concept
implementation
for
all
of
this,
but
but
the
idea
would
be
to
have
this
small
acceptable
steps
one
at
a
time.
So
first
one
would
be
just
the
CRI
API
and
and
the
code
annotations
to
get
get
started
and.
C
I
C
H
To
to
kind
of
spot
or
be
able
to
be
able
to
interpret
these
annotations
and
well
yeah,
and
then,
of
course,
the
feature
game
for
that,
but,
okay,
it's
it's
really
it's
really
minimal.
So
basically,
most
of
the
stuff
is
in
the
content,
would
be
the
content
around
times
the
CRI,
CRI,
API
and
and
then
the
kind
of
parsing
of
the
annotations
on
the
cable
itself.
But
that
would
be
basically
all
and
then
provision
unit
tests,
but
they
even
they
would
be.
I
G
It
will
work,
but
with
some
limited
approach,
so
using
this
using
this
approach,
workload
can
specify
what
it
requires
only
like
drum
memory
or
like
your
Ram
plus
some
other
combination,
but
you
will
not
be
able
to
provide
limits
per
particular
tier
of
memory.
However,
on
the
corner
side
view
anyway,
you
don't
have
any
any
variance
to
limit
all
of
those
so
as
a
first
step
to
enable
t-root
memory.
Yes,
it
will
have.
H
And
then
kind
of
well
we
have
not
two
practical
examples
this
already
for
for
cash
and
that
memory
bandwidth
and
block
IO,
but
but
I
guess.
If
we
get
get
this
supported,
then
we
have
kind
of
we're
gonna
explore.
For
example,
we
could
separate
some
some
overloaded
bits
from
that
Old
POS
class,
for
example
at
some
point,
possibly,
for
example,
have
some.
Oh,
oh.
H
D
C
Yeah
thanks
thanks
for
the
demo.
Yes,.
M
Thanks
for
the
demo
one
question
I
had:
if
I
can,
there
was
another
cup
I
think
proposed
a
while
ago,
specifically
for
qos
limiting
for
storage,
I,
think
it
was
gonna,
try
to
send
in
the
in
the
chat,
but
it
was
basically
proposing
directly,
including
in
the
Pod
spec,
like
the
number
of
iops
and
the
bandwidth
controls,
etc.
For
for
pods,
just
curious
what
your
thoughts
on
our
attack
cap
and
what
how
that
approach
kind
of
compares
to
the
class-based
system.
Here
absolutely.
H
H
I
mean
you
would
you
could
get
unexpected
results,
but
but
we
think
that
that
kind
of
this
would
simplify
the
user
interface
in
the
in
yourself,
so
that
the
cluster
administrator
that
could
like
have
even
per
node
configuration
so
for
in
node,
a
like
glass
Platinum,
wood
I
mean
have
totally
different
parameters
than
class
platinum
on
some
other
node,
depending
on
the
actual
Hardware.
That
is
on
the
Node.
G
One
more
point
to
that
is
what,
in
c
groups
the
limit
is
enforced
per
device,
so
putting
something
in
reports
back
it
exposes
to
the
user,
need
to
actually
know
which
particular
physical
device
will
be
present
on
and
off,
which
might
be
not
not
very
good.
Here's
is
exactly
simple
user
interface.
What
but
you're
specify
what
he
wants,
but
how
it's
implemented
it's
up
to
a
cluster
administrator
and
on
each
node
it
means
parameter
specific
to
web
node.
H
Upcoming
changes,
if
we
get
to
that
point
at
some
point
kind
of
having
this
access
or
permission
control,
then
you
can
also
limit
some
some
namespaces
or
or
users
from
really
using
the
kind
of
platinum
class.
For
example,
if
you
want,
if
you
don't
want
to
give
like
unlimited
access
to
this
bandwidth,
for
example,.
M
So,
to
kind
of
summarize
you're
saying,
like
the
other
Capcom
it's
on
the
user
side
versus
this
is
moving
the
cluster
administrator's
side,
and
this
would
be
done
something
for
example,
per
namespace.
So
the
when
you're
deploying
an
application,
depending
on
the
name
space
where
it
automatically
adds
some
annotation
or
something
like
that,
all
right.
That's
sort
of
the
thinking.
H
No,
not
no,
no,
not
not
really
so.
The
kind
of
future
steps
would
be
to
get
rid
of
these
annotations,
but
but
have
have
this
supported
in
Prospect.
So
these
are
good
specific
field,
important
to
specify
that
classes
for
users,
but
then
I
use
without
animation
that
you,
you
extend
the
resource,
both
the
mechanisms
as.
H
D
M
A
Yeah
I
think
one
one
of
the
trick
will
be
to
go
through
API
review,
because
my
usage
of
Border
annotations
are
not
ideal
and
I
think
it
will
be
at
least
frown
upon.
But
you
need
to
understand
because
it
can
go
through
API
review
at
all.
L
So
is
this
approach
kind
of
similar
to
like
the
priority
class,
where
you're
you're
saying
like
I
I,
want
high
priority
for
this
workload,
but
what
high
priority
means
depends
on
which
nodes
are
actually
in
your
cluster
but
like
from
an
application
developer
side
there?
Maybe
it's
a
separate
cap
but
like
you
might
want
to
say,
I
need
at
least
this
many
iops
and
I
want
the
scheduler
to
figure
out
where
to
put
this
foreign.
D
G
So
if
you
go
to
the
direction
of
scheduler
to
allocate
some
IO
Hawks
when
with
block
storage,
is
it's
almost
well,
it's
almost
impossible
to
guarantee
that
so
most
of
them
can
be
only
throttled,
but
but
this
cap
is
actually
going
towards
the
scenario
of
saying
what,
for
this
particular
resource,
I
want
with
priority.
G
So
like
we
have
two
example
like
memory
cache
like
cache
memory,
bandwidth
and
block
the
third
one
which
was
monal
was
mentioned,
like
memory
tuning
like
different
combination
of
memories,
what
can
be
available
and
when
workload
can
select
saying
what
like
I
might
require
slow
memory.
I
require
a
throttled
blocker
for
my
logger
container,
but
my
database
should
be
in
like
with
higher
memory
foot
like
Bend
this.
This
kind
of
bandwidth.
D
A
Yeah,
what
may
be
next
steps
to
agree
on
Direction
here
and
I
I
I'm,
not
sure
how
much
you
you've
done
with
scattering
part
of
it
and
like
I
mean
whenever
you
I
look
at
it.
A
It
feels
to
me
that
this
first
step
is
kind
of
simple
and
straightforward,
but
then
what
the
end
result
will
be
is
not
super
clear
and
yeah,
as
Bruno
said
like
it
doesn't
apply
to
memory
and
how
it
will
apply
to
Memories,
it's
kind
of
questions
arise
and
how
scheduler
will
decide
where
to
schedule
your
pod,
based
on
this
classes
from
so
maybe
this
may
help
yeah.
H
So
scheduler
knows
that
what
is
on
each
which
its
node
and
then
well
not
in
this
capital
in
the
future
steps,
and
then,
if,
if
a
certain
class
of
certain
resources
available
on
the
on
the
Node,
then
it
then
it's,
then
the
workload
is
scheduleable
there.
H
So
it's
it's
kind
of
really
straightforward
and
from
that
point
of
view
as
well
schedule
a
point
of
view
because
it
doesn't
it
it
only
kind
of
the
resources
from
the
scheduler
point
of
view
are
basically
just
key
value
Pairs
and
it
says
if
the
key
certain
key
with
certain
value
is
present.
I,
don't
know.
K
A
It
also
feels
like
get
on
time.
Class
of
sorts,
so
I
mean
all
sorts
of
runtime
hints
that
you
pass
to
I
mean
to
run
time
to
continue
runtime,
and
can
they
run
them
decides
what
to
do
with
that.
H
A
Yeah,
okay,
we're
almost
out
of
them,
but
my
other
question
will
be
like
how
much
you
like
you,
you
keep
everything
all
pack
and
maybe
One
Direction
is
to
make
it
super
opaque
and
just
say
that
these
are
hints
to
run
times
and
runtime
will
try
to
deal
with
that.
But
then,
at
some
point,
workload
may
not
be
portable,
because
runtimes
will
make
it
so
specific
that
they
only
know
how
to
deal
with
this
specific
classes.
A
I'm,
not
sure
how
much
you
want
to
expose
like
how
how
a
pack
want
to
do
it,
how
generic
want
to
do
it?
A
A
I,
also
see
a
couple:
announcements
of
caps,
phonetic
checkpoints
and
in
review.
Many
thanks
to
different
people
in
place.
Pod
vertical
skating
is
also
in
reviews
and
sandbox,
already
pod
conditions,
interview
and
waiting
for
direct
to
approve
and
finally
Peter.
There
is
a
topic
about
crying
catalogy
CV.
Is
there
something
you
can
cover
quickly
or
you
need
to
move
to
next
week.
J
Yeah
I
can
I
can
do
a
quick
overview.
This
is
just
there's
a
been
a
cve
that
unembargoed
yesterday
in
both
Kyle
and
containerdy,
about
exact
sync
requests
over
basically
using
an
unbounded
amount
of
memory
and
I
wanted.
So
we
both
of
the
implementations,
came
to
a
conclusion
that
we
would
just
cast
the
amount
of
memory
used
by
an
exact
sync
request
or
like
Pro,
by
processing
the
output
to
16,
Megs
and
so
I
just
wanted
to
bring
the
circle
back
to
everyone.
J
Now
that
it's
been
an
embargoed
and
patched,
we
should
probably
document
that
behavior
or
think
of
a
more
official
way
of
proclaiming
that
limitation,
so
that
I
know
in
the
docker
shim.
It
was
just
passed
in
the
same
way.
So
that's
how
we
kind
of
call
us
to
the
16
Megs.
J
That's
a
good
idea
and
and
I
can
take
an
action
item
to
to
do
that.
Documentation.
J
Well,
yeah
I
just
wanted
to
check
in
about
that
and
see
if
anyone
had
any
thoughts
or
you
know,
follow-up
work
that
they
wanted
done
to
that.
A
Thank
you
for
bringing
it
in.
Can
you
link
post
a
link
into
CCB
in
agenda
items,
so
people
will
be
able
to
look
at
it.
A
A
N
Yeah
Sergey
I
think
I
have
posted
one
in
the
chat.
Basically,
this
is
a
request
for
the
kept
Revival
which
which
got
closed
automatically.
You
know.
Basically,
you
know
we
are
facing
an
issue
with
the
container
d.
N
You
know
container
where,
when
we
exec
right,
there
is
no
user
option.
I
think
this
is
an
old
topic.
I
think
some
of
the
guys
here
might
be
aware,
including
I,
think
Kunal
yeah,
so
yeah
I.
C
C
This
create
a
new
cap
in
the
new
format
and
start
from
there
like
just
given
where
we
are
I'm,
not
sure
if
we
can
get
this
in
125,
but
that's
where
I
would
start
like
create
a
new
cap.
Like
add
Mike,
Brown
me
and
other
folks
interested
here
as
reviewers
and
approvers
and
we'll
be.
We
can
take
it
from
there.
I
guess
like.
J
K
C
N
Yeah
right
I
found
that
original
author
is
no
more
active.
I
think
it
looks
like
he
has
left
so
fine,
then
I'll
create
a
cap
in
the
new
format
and
maybe
I
need
help,
because
I
am
not
continually
expert
myself
because
I'm,
just
an
user
of
containerdy
in
my
code,
but
yeah
I,
will
need
help
from
you.
Guys
to
you
know,
guide
me
through
that.
Yeah.
A
Thank
you,
bye-bye,
everybody,
so
good
rest
of
your
day
right.