►
From YouTube: Kubernetes SIG Node CI 20230104
Description
SIG Node CI weekly meeting. Agenda and notes: https://docs.google.com/document/d/1fb-ugvgdSVIkkuJ388_nhp2pBTy_4HEVg5848Xy7n5U/edit#heading=h.2v8vzknys4nk
GMT20230104-180415_Recording_1692x1342
A
Hello,
hello,
it's
Wednesday,
January
4th
2023
and
it's
signal
CI
meeting
welcome.
Everybody
today
is
the
first
meeting
after
holidays,
so
not
many
people
and
not
much
on
agenda
I
would
want
to
consider
it
on
triage
and
but
first
I
wanted
to
point
this
out
that
after
holidays
I
looked
at
test
grid
and
this
cryo,
all
the
characters
are
suddenly
failing.
Like
totally
failings.
It's
not
even
starting
yeah.
B
In
the
cryo
world
we
have
we're
aware
of
this
and
we're
working
on
it.
It
was
a
merge
that
we
made
before
the
break
so
should
be
fixed
soon.
C
Yeah
we
are
this
morning,
Ramesh
told
me
that
he
was
able
to
run
it
successfully
locally
and
we
just
have
to
bump
the
cryo
version
in
test
info.
A
A
Okay,
so
yeah
I'm
glad
it's
been
attacked
so
yeah
and
let's
go
to
the
port,
not
many
items
there.
Even
though
there
were
holidays
something
new
appeared
here.
A
A
Yeah,
it's
mostly
storage,
I'm,
not
sure
how
much
they
need
from
us.
A
A
A
A
Yeah
go
ahead
open.
A
D
A
It's
it's
original
author,
okay,.
A
A
A
We
often
assign
it
to
Daniela,
but.
A
B
I
could
try
but
I.
You
know,
I,
don't
really
know
much
about
Windows,
but
I
could
do
a
cursory
review.
A
A
Now,
let
me
ask
around:
maybe
I
can
find
somebody
who
who
knows
Windows
who
can
help
everybody.
You
will
spend
time
on
some
other
people,
maybe
use
it
okay
with
you,
but
you
still
want
to
think
yeah
yeah,
that's
fine,
okay,
I
will
I
will
ask
it
out.
A
Let's
just
use
printf
so
forever,
config
too
yeah
same
it's.
Some
big
change
across
multiple
seeks.
Don't
need
to
take
a
look
sick
testing,
typically
I,
look
into
those
foreign.
A
Yeah
I
also
yeah.
There
were
some
questions
about
some
LG
TMT
PRS,
but
it's
for
Francesca.
We
want
Francesco
to
take
a
look,
so
I
will
GM
if
his
background
location
or
holidays.
A
Excuse
me
yeah:
if
nothing
about
test
I
think
we
can
go
to
bhaktiash
I
want
to
finish
with
all
the
triage
and
then
we
can
discuss.
Maybe
some
active
tasks
that
you
have
black
triage
for
battery
to
go
here.
A
A
Every
time
after
holidays
or
occasions,
it's
growing
very
big.
A
A
A
A
Yeah,
it's
clearly
stating
that
entire
Port
needs
to
be
scheduled,
I
think
it's
missing
the
saying
of
how
it
works.
A
B
B
I
yeah
sure
you
can
add
needs
information.
It
looks
like
it
actually
looks
like
Adrian's
already
pretty
much
done.
The
triage
part
and
they're
just
trying
to
figure
out
the
two
of
them.
A
A
A
A
A
I'm
sorry
I
forgot.
You
h,
h
e
r.
A
Let's
use
that
forward,
but
yeah,
if
you
can
help
with
that
great.
B
It
might
be
for
quotas
yeah
well
for,
like
storage
with
Emerald
storage
quota.
What
run
time
are
they
using.
B
F
Yeah,
it's
an
alpha
feature
guide
for
kilowatt,
it's
all
mer
and
all
of
my
list
to
look
into
deprecating
or
keeping
this
feature
gate
around.
A
A
And
can
be
reproduced
for
the
screen
removed,
okay,
so
yeah.
It
seems
that
it
have
plenty
details:
okay,.
F
I
think
there's
too
many
bugs
to
promote
it
to
Beta,
but
it's
the
evaluation
that
we
need
to
do.
B
I
think
I,
remember,
hearing
also
that
there
was
a
like
bug
in
projects
in
general,
where,
like
a
user
inside
of
container,
could
change
the
project
that
they
have,
so
it
doesn't
even
actually
work
to
do
secure
like
isolation
for
security,
for
like
from
a
particularity
standpoint,
so
I
suspect
we
may
actually
be
looking
at
deprecating
it
instead.
E
A
Yeah
I
I,
don't
really
know
this
code
very
well.
To
tell
whether
this
description
description
sounds
logical.
If
you
just
read
it
but
I
I,
don't
look
at
this
chord.
A
A
A
A
D
D
A
A
A
B
B
B
A
B
A
A
Sweetie
but
a
cute
chick
name
for
download.ct.
B
A
A
A
A
A
A
A
I
know
that
in
Red
Hat
you
use
a
lot
of
long
terminating
like
nodes.
Is
there?
Is
this
probably
my
record?
Is
there
something
you
want
to
look
at?
A
The
problem
is
that
when
node
is
gracefully
being
terminated,
some
Demon
said
that
schedule
like
the
tolerate
everything,
because
they
want
to
schedule
ports
very
early
in
the
not
life
cycle,
so
not
just
bootstrapping,
and
they
immediately
want
something
on
this
note,
so
they
tolerate
everything
they
want
to
make
sure
that,
even
though
they
tolerate
bootstrap
phase,
they
don't
tolerate
I
mean
they
don't
been
scheduled
on
graceful
shutdown
stage.
F
A
A
D
D
A
A
A
A
D
A
Yeah
you
mentioned
that
it
may
already
be
done
so
yeah
we'll
be
good.
B
A
D
A
A
I
want
to
take
a
look
at
this
piece
of
code
and
add
API
version.
A
A
A
Specific
placing
code
so
probably
easy
to
find.
A
A
So
this
issue
was
about
Google
being
restarted
and
then
containers
are
shown
as
not
ready
during
this
restart
and
I.
Think
we
have
this
reported
many
times
already.
Could.
A
I
think
there
was
an
idea
to
fix
it
by
checking
API
server
for
last
status
of
the
spot.
I
mean
Google's
been
restarted,
so
it
doesn't
store
a
Readiness
state
in
any
place
and
since
it
doesn't
store
it
when
it
starts,
it
assumes
that
pod
is
not
ready.
A
So
if
Kubota
restarts-
and
you
have
like
many
ports,
all
the
ports
will
go
to
not
ready
first
and
then
to
ready
back
to
ready.
So
it
caused
a
lot
of
problems
with
like
let's
say
you
have
endpoints
configured
for
all
the
sports,
and
this
endpoint
collection
will
be
updated
once
and
then
updated
another
time
when
they
get
to
ready
State
again
so
yeah.
A
Let's
I
mean
it's
not
very
good
behavior
it
just
all
we
can
Implement
and
we
don't
have
really
good
idea
how
to
store
a
Radiance
status
between
restarts.
That's
why
it's
still
here.