►
From YouTube: Kubernetes SIG Node CI 20220615
Description
SIG Node CI weekly meeting. Agenda and notes: https://docs.google.com/document/d/1fb-ugvgdSVIkkuJ388_nhp2pBTy_4HEVg5848Xy7n5U/edit#heading=h.2v8vzknys4nk
GMT20220615-171830_Recording_1480x1032
A
Hello,
hello,
it's
June,
15th,
already
2022,
it's
a
signaled
CI
meeting,
welcome
everybody.
We
don't
have
too
many
people
today.
Let's
just
go
through
I
mean
first
is
a
small
announcement
from
my
side.
I
just
had
a
baby,
a
few
like
one
months
ago
and
I'm
going
to
parental
leave.
Now
so
I
will
be
unavailable
for
two
months.
Yeah
I'll
be
back
in
September.
A
So
next
I
think
is
triage
unless
anybody
has
specific
topics.
Brian
typically
we'll
just
introduce
some
like
I
mean.
If
you
want
to
discuss
some
specific
flakes,
you
can
put
them
in
another
agenda
and
we
can
discuss
them.
Otherwise,
you
just
go
through
triage
and
look
at
what's
happening.
A
Sorry
I
have
trouble
getting
the
shot.
Yeah
shot.
Is
here
yeah?
It's
something
invite!
Oh
you
don't
see
my
screen
yet.
Let
me
share.
A
Okay,
yeah:
let's
go
through
Josh.
A
A
Okay,
I
will
put
in
needs
author
attention
original
color.
A
Typically,
this
kind
of
prcm
easy
to
get
started
with.
A
A
Some
more
config
and
featured
security,
so
this
is
not
a
test
specific
I
wonder
how
many
tests
were
added.
If
we
need
to
look
at
it.
C
A
Okay,
I
think
we're
done
with
this
yeah
we're
done
with
this.
Does
anybody
wants
to
discuss
any
ongoing
issues?
Do
you
have
any
troubles
is
what
you're
working
on
I
looked
at
test
Grid,
it's
mostly
the
same
as
last
time.
I
didn't
update
this,
but
yeah.
It
was
mostly
the
same
so.
B
Seriously,
I
got
a
input
about
the
test
grip.
This
is
something
that
you
told
me
to
check
at.
There
were
some
flaky
chests
one
of
them
involving
open,
API,
failing
on
a
call
about
getting
a
pod
status,
so
I
I,
measured
the
guy
who
had
a
a
PR
out
to
fix
it,
salty,
sh
or
whatever.
Anyway,
he
pushed
out
a
fix
for
it.
B
B
A
If
it's
still
open,
then
we
we
can
just
highlight
like
I
mean
you
can
comment
on
this
issue,
saying
like
it's,
so
it's
still
red,
so
people
wouldn't
close
it
mistakenly.
D
There's
actually
an
issue
that
I
have
been
working
on
and
I'm
actually
struggling
with
it.
I
was
wondering
if
anyone
would
be
able
to
help
me
out,
so
it's
related
to
introducing
a
repo,
a
staging
repo
in
kubernetes,
so
I
I'll
just
share
a
PR
actually
on
the
channel
to
find
it
and
essentially
I,
think
I
started
off
initially
by
proposing
the
entire
PR
and
then
try
to
kind
of
downsize
it
by
just
introducing
the
bare
minimum.
D
You
know
having
the
owners
the
like
the
readme
and
things
like
that,
but
the
CI
is
just
not
happy
and
I'm
struggling
to
have
the
dependency
resolution
as
well,
because
I
think
in
staging
they.
If
there's
a
strange
dependency
and
the
bot
populates
all
the
components
of
the
staging
repo.
So
I
was
wondering
if
anybody
has
experience
in
this
area
and
would
be
able
to
point
me
to
the
right
direction.
E
D
So
I
refer
to
like
this
in
the
community
repo,
there
is
a
document
which
talks
about
how
do
you
create
a
staging
repo
per
se
and
I
followed
all
the
steps
and
that
where
that's
where
it
was
mentioned,
that
you
create
a
DOT
GitHub
directory?
And
you
populate
all
these
things,
but
I
think
one
of
my
major
issues
is
kind
of
dependency
resolution
because,
like
when
I
do
a
vendor
update
it's
not
getting
essentially
I'm
getting
errors
and
when
I
don't
do
it.
D
The
some
like
most
of
the
tests
pass
and
and
a
lot
of
dependency
vendor
dependent
repos,
are
getting
delayed
deleted
in
the
vendor
itself.
So
I
just
I
just
stuck
with
this
for
about
a
week,
probably
more
than
that
and
I
just
don't
know
what
to
do.
A
So
I
think
the
best
place
for
that
is
cigarch
called
code
structure
or
like
what
is
quote.
What
is
a
sub
project
for
sea
garage
about
our
code
is
laid
out.
Okay,
it
may
be
a
good
place
to
get
information,
but
did
you
find
any
other
like
I
mean
I
would
go
with
like
finding
another
PR
that
doing
similar
thing
like
moving
something
yeah.
D
So
I
I
tried
actually
I
did
I
tried
that
as
well,
but
like
I
wasn't
able
to
find
proper
and
analogy.
I
think
it
was
like
the
last
contribution
to
Stage
repo
was
a
while
ago,
and
a
lot
has
changed
since
then,
so,
like
I
think
they
were
there's
basil
and
things
like
that
previously
and
in
fact,
I
I
had
a
PR
which,
which
was
passing
fine
and
working
previously,
but
I
just
can't
get
my
head
around
it
anymore.
A
D
A
A
Sure
sorry,
it's
hard
to
find
out
really
quickly.
D
D
So
if
you
look
at
the
issues
per
se,
so
I
have
actually
another
one.
There
are
two
repositories
in
kubernetes
sick
that
rely
on
this
API.
So
this
is
I,
think
the
main
issue
and
it
was
essentially
like
we
had
previously
planned
to
introduce
this
repo
in
kubernetes,
and
so
this
was
the
issue
and
I
will
actually
share
another
eventually
fixed
this
one.
D
Yet
this
is
sorry
that
was
the
TR,
and
this
was
the
issue
and
the
goal
was-
and
this
was
pointed
out-
I
think
by
dims-
and
he
said
that.
Okay,
we
have
this
API,
which
is
actually
sitting
externally
in
an
external
GitHub
organization,
and
there
are
two
repositories
that
are
depending
on
this
external
API,
which
is
known,
feature
Discovery
and
scheduler
plugins,
both
of
which
sit
in
kubernetes
organization,
so
that
was
kind
of
where
this
kind
of
re-cropped,
because
I
our
initial
goal
was
to
propose
it
in
kubernetes
and
initially
it
was
rejected.
A
Nothing
yeah
architecture,
code
organization
is
the
best
place
to
like
they
discussed
exactly
these
kind
of
questions.
A
Okay,
any
other
comments:
yeah
I
I
personally
looked
at
like
one
of
the
failing
tests
is
performance
tests
and
I
fixed
it
before
so
now.
My
struggle
is
to
reproduce
it
locally,
because
I
mean
the
problem
with
the
test
is
basically
it
creates
a
I
mean.
This
is
extremely
simple:
it
creates
a
port
and
it
waits
for
Port
to
complete
and
then
measure
time
of
this
execution.
So
Port
runs
for
me
perfectly
fine,
but
it
doesn't
run
good
and
like
kubernetes
environment,
and
this
port
is
massive.
A
It
uses
some
tensorflow
calculations,
so
yeah
and
log
output
is
huge,
so
I'm
struggling
to
understand
what
is
strong
in
test
environment
so
yeah.
This
is
my
struggle
and
I'm.
Looking
at
this
right
now.
So
I
don't
know
if
the
description
sparked
any
ideas,
but
my
current
thinking
is
somebody
whom
kills
the
port,
because
I
don't
see
any
other
reasons
for
Port
to
disappear.
I
like
to
fail-
and
this
is
something
I
will
be
looking
into
somehow
I
mean
I'm.
A
A
E
E
C
A
Okay,
seven
bucks
attached
nothing
happens
this
time,
because
it's
this
everything
is
quite
calm.
Oh
I
think
everybody's
eating
caps
right
now,
so
I
interviewed
a
couple
caps.
Yesterday,
it's
very
time
consuming
it's
Eurasian
and
like
you're,
trying
to
understand
all
the
details
and
comment
on
design
which
is
hard,
I,
think
everybody's
busy.
With
that
these
days.
F
I
took
a
look
into
this,
so
I'm
not
really
sure
if
this
is
and
I
I
left
like
a
detailed
comment
at
the
bottom.
If
this
is
the
the
one
I'm
thinking
of,
but
essentially
it
doesn't
seem
like
there's
a
way
to
avoid
this.
You
know
super
Superfluous
right
header,
like
warning
without
re-architecting
how
the
code
Works
so
I,
don't
know
if
this
is
like
a
bug
or
or
more
of
like
yeah,
the
way
that
we
structured
this
git
kublic
container
logs.
F
F
Yeah
so
like
that,
400
response,
like
I,
would
expect
a
400
to
mean
something's
wrong
with
your
request,
but
every
from
the
logs
provided
the
request
was
valid,
but
the
command
used,
a
timeout
which
canceled
the
canceled.
The
logs
request
after
like
two
seconds
and
I,
think
the
400
might
mean
that
the
connection
was
like
terminated
by
the
client
but
reading
through
this
restful
library
that
we
use
to
serve
like
the
logs
requests
in
kublet.
I
I
couldn't
find
anything
that
says
that
explicitly
so
I
I
can't
say
for
sure.
F
A
B
A
B
It
big
enough
yeah,
it's
good
enough.
A
D
A
A
A
A
I
mean
you
can
try
to
reoperate
and
see.
What's
going
on,
I
mean
just
a
different
bug
and
like
trying
to
follow
steps,
typically
a
good
way
to
double
check
that
what.
A
Is
actually
what's
happening,
so
this
is
valuable
contribution
because
sometimes
people
just
misconfigure
things.
E
B
B
A
We're
done
with
the
battery.
Is
there
anything
else
anybody
wants
to
discuss
today.
B
B
A
A
But
this
may
be
a
need.
I
know
it's
a
build
number
because
yeah,
it's
increments
too
slowly.
Maybe
this
is
it
I
mean
I,
always
thought
I'd,
say
testing
for
comments.
B
A
Typically,
if
it's
in
mystery
side
here
so
I
I,
don't
question
myself
too
often
on
that.
So.