►
From YouTube: Kubernetes SIG Node CI 20230531
Description
SIG Node CI weekly meeting. Agenda and notes: https://docs.google.com/document/d/1fb-ugvgdSVIkkuJ388_nhp2pBTy_4HEVg5848Xy7n5U/edit#heading=h.2v8vzknys4nk
GMT20230531-170533_Recording_2086x1340.mp4
A
A
Today
on
agenda,
we
have
a
few
updates
on
there's
that
we
extended
test
coverage,
so
Standalone
mode
tests
are
green.
Now
it
took
us
couple
iterations,
you
may
notice.
Last
time
there
were
some
problems
and
now
these
problems
are
gone.
So
we
have
all
the
test.
Executions
successfully.
A
To
24
hours,
so
this
is
all
Alpha,
and
this
is
regular.
One
thing
we
may
want
to
do
here
is
to
extend
the
test
coverage,
but
today
is
the
only
test
we
run.
Is
we
create
a
static
port
and
satisfied
with
that?
We
may
need
to
add
the
upgrade
scenarios,
maybe
a
static
port
with
some
storage.
A
Maybe
some
security
features
as
well,
just
to
name
a
few
but
I
think
over
time.
It
will
keep
extending
this.
At
least
you
have
a
basis
for
that,
and
now
you
can
look
further
and
I
hope
it
will
catch
situations
like
we
had
before.
A
Like
last
released,
we
had
two
caps
that
broke
Standalone
mode.
Now
we
should
be
able
to
catch
it
very
quickly,
yeah,
okay,
next
one
is
arm64
I'm
64
quote
for
that
was
merged.
Test
grid
definition
was
merged.
Now
the
problem
still
persist
after
the
first
verse.
There
was
a
second
verse,
some
adjustments
made,
but
we
still
have
it
failing
you
see.
Zero
can
start.
A
So
this
is
something
that
needs
to
be
investigated
to
try
to
pick
people,
but
hopefully
we
can
have
this
address
soon
and
in
128
you
have
rm64
tests
available.
B
A
B
I
I
believe
I
had
this
was
done
actually
as
part
of
topology
manager
GA
graduation.
But
if
you
want
to
move
that
to
periodics,
maybe
we
can
create
a
separate
bug
just
for
that,
because
I
like
that.
That
issue
is
completely
closed
now,
because
topology
manager
graduated
to
ga.
A
Yeah
the
problem
with
CI
jobs
that
are
not
periodically
running
and
we
may
miss
some
issues
by
the
way.
I
still
don't
understand
the
reason
why
we
don't
run
optional
CI
jobs
periodically,
it
will
probably
would
be
a
good
idea.
You
can
even
do
it
through,
like
dummy
PRS.
A
When
we
have
mtpr
that
just
test
every
job
but
yeah,
maybe
I
keep
it
in
sync:
periodics
and
CI.
Jobs
is
not
sometimes
say
going
away
from
each
other
and
like
Ci,
job
is
not
fully
covering
with
our
periodics
running
or
other
ways
around,
and
it
makes
it
harder
to
understand
how
to
test
the
pr
and
what
kind
of
tests
are
available.
C
B
No
I
was
kind
of
thinking
about
this,
but,
like
my
understanding
was
that
we
have
periodics
and
then
we
have
pre-submits
and
the
purpose
is
kind
of
different
from
each
other.
In
the
sense
that
you
know
you
submit
your
pre-submit,
you
can
enable
a
job
at
a
time
of
a
PR
for
a
very
specific
case,
and
and
that's
why
we
have
them
as
optional,
so
enabling
them
as
part
of
periodics
might
be
just
because
of
resource
constraints.
I
believe
I
kind
of
want
to
have
that
separation.
And
but
that's
just
my
understanding,.
A
Yeah
so
yeah,
the
problem
is
that
approaches,
please
submits,
are
degrading
over
time
and
police
admits
also
diverging
from
periodic's
definition.
So
if
there
are
some
Alpha
feature
flag
or
like
some
feature
flags
that
was
defined
on
periodics,
it's
not
on,
please
submits
any
longer,
and
please
admits
wouldn't
run
that's
kind
of
issues
happening
all
the
time
right.
That's
that's.
Probably
the
reason.
A
Yeah
and
probably
not
signaled
would
be
doing
it.
I
mean
what
you
can
do
is
to
just
run
all
the
signal
related,
please
admits
periodically
with
mcpr,
but
I'm,
not
sure
if
it's
it's
a
good
practice
mode.
D
Okay
and
I
will
probably
reach
out
for
to
seek
testing
if
one
one
to
as
part
of
the
investigation
for
this
job
I
mean
alongside
not
because
of
so
I
will
ask
if
there
are
practice
or
blocks
I
can
report
here.
If
so,
if
they're
established
approach.
E
So
I
have
shared
the
first
rock
for
the
guidance
around
test
coverage.
We
spoke
about
it
like
a
few
weeks
ago
and
let
me
see
how
to
have
Public
Access.
E
E
E
E
A
Okay,
so
it's
basically
the
same
as
we
discussed
before
on
one
of
the
previous
meetings.
A
E
Yeah
I
just
went
ahead
and
granted
common
comment
permissions
to
the
group,
so
it
will
be
really
creative.
You
can
look
at
it
and
drop
in
some
feedback
around
structure
or
around
any
of
the
guidance
that's
provided
at
the
top.
A
A
C
C
A
I
think
dixio
was
sharing
same
document
before
and
the
adjustment
was
made
like
cryo
was
added
and
something
else
yeah
anyway,
please
review
and
I
think
we
can
start
applying
this
classification
of
tests.
A
Okay,
let's
go
next.
A
This
is
okay,
let's
go
so
performance
really
quickly
and
then
dashboards.
A
A
A
B
A
A
A
Yeah
it
needs
to
anybody's
interested
to
the
users,
seems
to
be
quite
straightforward.
A
A
Yeah,
it
looks
like
a
somebody's
condition.
D
A
So
yeah
I
wonder
what
was
the
sport
actually
run
to
completion
then
after
Google
restart
it
would
fail
to
tell
that
it's
completed.
E
D
E
D
C
C
A
Okay,
yeah
I'll
fix
it
later.
A
Multiplication
notches
for
scale
Docker,
CLI.
E
A
Yeah
but
I
think
it
doesn't
clean
up
because
of
some
duplication
like
this
errors
happening
some
ports.
Duplications
I
think.
A
This
is
PR,
that
is,
reference
here
is
for
project
application.
C
A
A
Because
it's
it's
expected
question.
A
Okay,
so
they're
talking
about
deployment
Mark
searching
much
available
from
very
first
glance,
is
not
signaled
related.
A
Okay,
reapply
yeah
I
will
make
it
six
UI.
A
Okay
yeah,
so
we're
done
with
three
Azure
box:
I,
don't
have
much
energy
in
me
because
of
ways
to
go
through
need
information.
If
somebody
wants
to
take
over
and
get
some
review
here,
if
not,
we
can
just
finish
early
today.