►
From YouTube: Kubernetes SIG Node CI 20230705
Description
SIG Node CI weekly meeting. Agenda and notes: https://docs.google.com/document/d/1fb-ugvgdSVIkkuJ388_nhp2pBTy_4HEVg5848Xy7n5U/edit#heading=h.2v8vzknys4nk
GMT20230705-170603_Recording_1920x1020.mp4
A
C
B
E
B
A
A
B
A
A
A
D
A
E
A
C
C
F
E
D
A
C
A
Okay,
so
we
feel
some
fixed
and
we
still
have
work
on
others
yeah.
If
you're
interested,
please
jump
in
and
I
wanted
to
check
with
what
is
not
here.
Yeah
we
have.
We
still
have
this
topology
management.
All
other
tests
in
I
mean
CI
test
seems
to
be
working,
but
a
regular
periodics
are
not
so
we
need
to
fix
previous
at
some
point.
F
It's
not
an
agenda
item,
but
I
just
want
to
quickly
mention
this
someone
and
signaled
the
channel
asking
about
how
to
run
the
end-to-end
Cuba
test
and
if
I'm
not
mistaken,
we
do
have
the
docs
for
that
and
if
you
can
remind
me
where
the
link
is
I
can
Shadow.
If
you
want
to
share
and
Signal
it's
the
latest
thread,
it
could
not
be
helpful.
F
A
E
F
E
D
A
E
E
G
A
C
A
A
E
A
Send
to
me
if
anybody
wants
to
take
a
look,
please
welcome.
So
what
the
issue
was
is
if
Cobalt
is
restarting
while
init
container
still
executing,
we
can
lose
some
State
and
I.
Think
this
test
will
test.
That
state
is
not
actually
being
lost.
We
found
that
during
the
pr
poor
request
review
for
sidecar
containers,
cap.
A
A
A
A
F
A
F
E
F
G
A
A
A
C
A
A
Oud
yeah:
this
is
what
we
discussed
already.
We
just
moved
into
in
progress
because
talk
is
working
on
that
and
we're
done
with
this
part.
Okay,
we
spent
half
a
meeting
called
test
related
things
and
cleaned
up
any
comments.
Feedback
I
want
to
get
into
bugs
because
we
have
the
long
time
with
no
meetings
and
I
want
to
make
sure
that
we
covered
everything.
A
B
A
A
B
Yeah,
it
seems
like
they
have
they
have
this
pod.
That's
for!
Oh,
maybe
it's
a
PSA
where
those
debug,
but
it
doesn't
have
a
capability
that,
like
you
know,
because
this
container
D
containerdy
gets
net
raw
by
default.
The
crowd
doesn't,
and
so
a
lot
of
people
assume
that
with
the
default
capability
set
you'll
get
in
that
raw.
So
that's
not
necessarily
the
case.
B
A
E
A
B
I
think
there
is
something
to
be
said
about
having
a
base
template
of
the
excuse.
Wicked,
big
but
I
think
that
the
drop-in
configuration
where
we'll
make
a
note
IP
make
more
sense.
Because
then
you
could
have
a
piece
of
drop-in.
You
could
say:
that's
that's
positive
mode.
I
see
that's
different
for
each
node,
so
I
would
actually
say.
Maybe
that
393
is
a
precursor
to
this,
but
I
don't
think
we
should
do
it
until
we
have
the
drop-in
configuration.
E
B
A
B
Interesting
yeah
I,
don't
I
I,
don't
think
that
that
would
be
terrible.
I
mean
it
kind
of
depends
on
how
we
implement
it.
The
way
cryo
did
it
is
it
basically
has
one
function
for
updating
a
file,
and
then
it
like
recursively
calls
that,
but
it
is
aware
of
what
the
first
file
is
versus.
What
the
subsequent
ones
are
so
easy
to
consider.
So.
F
B
You
know
yeah,
it
I
understand
why
one
might
want
to
set
that
intent
of.
Like
you
know,
node
IP
can
only
be
set
in
the
drop-in,
but
that
kind
of
feels
like
adding
a
configuration
option,
that'll
kind
of
it's
like
kind
of
narrowly
scope,
because
it's
still
up
to
the
cluster
admin
to
know
what
they're
doing
so,
it's
not
like.
We
can
really
prevent
a
user
from
shooting
themselves
in
the
foot
because
they
could
just
set
the
same
node
IP
in
each
of
those
files.
E
B
Yeah
I
don't
know
if
we
have
any
sort
of
compatibility
or
checks
and
more
reference
to
cry.
Cryo
has
automated
checks
to
make
sure
that
all
of
the
configuration
fields
are
present
scli5
to
vice
versa,
with
the
exception
of
the
couples
you
know
ones
where
it
makes
sense
not
to
but
I,
don't
know
if
we
have
ever
thought
of
having
that
kind
of
compatibility
of
a
one-to-one
relationship
between
the
two.
A
Yeah,
what's
confusion
here?
Is
it
really
good?
Yeah
was
I,
saying
like
based
on
description
of
boxes
and
even
after
they
try
to
do
is
through
command
line
arguments.
It
still
doesn't
work
and
I'm,
not
sure
what
doesn't
work
is
it
did
they
removed
from
config
and
only
used
command
line
or
the
remote
box.