►
From YouTube: 20180911 sig cluster lifecycle
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
B
Well,
I
just
said
it
one
item:
it's
about
the
kubernetes
anywhere
change,
I
wanted
unblocked
and
reviewed
so
that
we
can
merge
it
today,
like
Illya
added
a
comment
that
we
need.
Docs
for
this
and
I
kind
of
disagree,
see
what
basically
wants
to
document
how
to
download
the
latest
correct,
we've
and
I
think
that
we
shouldn't
document
anything
there.
B
A
B
B
C
It's
like
the
next
thing
on
the
agenda
was:
let's
fix
the
alcohols
to
that
repo,
because,
as
that
repo
is
now
basically
just
being
used
as
an
end-to-end
runner
for
cube
admin,
and
we
don't
want
to
use
it
for
anything
else.
The
people
that
are
our
debugging
it
and
running
you
know
maintaining
those
tests
for
cube.
Admin
need
to
just
have
access
to
it
directly.
I
think
I
mentioned
before
that.
I
could
do
that,
and
I
haven't
figured
out
how,
if
anybody
knows
how
those
Eckles
are
set
up.
Otherwise,
I'll
go
start
poking
at
github.
B
B
B
A
B
So
when
I,
when
I
started,
looking
at
the
failures,
it
was
about
the
change
from
Peter
that
he
removed
the
deprecated
ignore
preflight
checks.
Now
it's
called
ignore
pre-flight
errors
and
I
went
back
in
history
and
I
couldn't
find
a
merge
that
emerged
in
the
mail
repo,
an
emerging
testing
for
that
made
sense
to
like
break
this
outside
of
our
like
failure
with
the
deprecated
and
then
I
just
went
and
renamed
the
job.
And
now
the
history
is
lost,
lost.
B
A
It
between
your
anecdotal
evidence
and
my
anecdotal
evidence.
We
have
two
signs
pointing
to
the
same
area,
which
is
that
the
version
of
we've
was
not
running
on
the
machines
and
when
you,
when
I,
was
looking
in
debugging
stuff
from
the
logs
which,
by
the
way,
there's
there's
a
much
frustration
on
my
side
with
the
ability
to
debug
from
the
logs
like
when
we
built
sonobuoy
for
his
respective
detection.
We
grab
all
the
pod
logs
as
part
of
that
too,
and
we
don't
actually
have
the
bad
logs,
because
we
had
that.
A
B
A
We
knew
certain
order
of
problems
that
occurred
right
and
just
by
bisecting
that
quarter,
but
yeah
there's
not
enough
data
there.
That's
just
a
general
problem
is
that
kubernetes
anywhere
is
really
hard
to
debug
like
super
hard
to
debug,
and
then
it
also
wraps-
and
you
don't
see,
a
bunch
of
commands
and
everything
else
so
like
you,
because
that
execution
is
totally
hidden
to
you.
You
don't
see
the
lower
level
details
going
on,
but
that's
only
part
of
the
problem.
A
A
A
Call
I
might
even
like,
because
coop
kins
is
a
monster
we
have
weirdness
I
might
just
talk
with
tested
for
the
trauma
fix
this
problem,
but
yeah
Lord
use
federated
testing,
listen
a
boy
because
we
are
in
the
other
own
testing
infrastructure
built
around
suitably
anyways.
First
stuff
might
just
automate
that
stuff
to
plug
in
the
federated
testing
with
the
cluster
API
and
could
be
an
admin
just
be
done
with
it.
A
B
C
B
B
B
Alexander
somebody
submitted
that
that's
like
a
that's
like
a
separate
agenda
item.
Somebody
submitted
an
issue
at
the
cube
ad
every
boy
and
should
they
get
out
that
he
claims
that
we
should
support
basic
authentication
via
cube.
Atm
join
and
I
saw
historical
issue
about
this
yeah,
but
I
wanted
your
opinion
on
this
because
I
think.
B
Basically,
the
user
is
saying
that
he
is
adding
the
basic
authentication
file
as
a
volume
for
the
API
server,
but
then,
when
he
tries
to
keep
ad
and
join
with
the
token
he
wants
to
provide
the
user
and
password
for
the
basic
authentication.
Otherwise
he
is
claiming
that
this
fails
for
him.
Like
does
anybody
have
comments.
B
B
D
C
B
C
A
B
A
F
C
C
F
A
B
A
A
Chuck
and
Jason
have
updated
the
H,
a
installation
instruction
so
more
eyes
to
evaluate
that
would
be
helpful.
Those
installation
instructions
should
be
leveraging
the
new
control
plane
join
features.
So
if
folks
have
time
to
verify
those
PRS
independently,
because
last
time
the
documentation
was
updated.
But
there
was
a
tiny,
tiny
little
typo
and
we
caught
we
caught
Twitter
flack
because
of
the
typo,
because
somebody
actually
had
a
problem
trying
to
do
their
H
a
deployment,
and
then
it
was
only
because
of
the
way
kabane
IAM
doesn't
have.
A
Doesn't
it
has
an
omit,
empty
sort
of
silent
failure
and
the
api
machinery
stuff
and
because
there
was
a
minor
typo,
it
wasn't
caught
until
after
the
fact.
So
it
having
people
to
go
through
the
entire
instruction
set
for
setting
up
and
deploying
an
hae
cluster
using
the
updated
instructions,
I
think,
would
be
beneficial.
A
A
The
that's
the
list
that
we
have
tried
patching
the
skip
list
for,
and
we
can't
do
you
remember
that
from
last
cycle
there
there
are
a
set
of
tests
that
will
always
fail
on
the
on
the
upgrade
portion,
just
because
of
something
I
like
Lucas
and
I
tried,
Lucas,
Jason
and
I
tried
to
fix
it
and
test
infra.
There's
a
bug
somewhere
in
testing
from
that
prevented
us
from
actually
they'd
ignored
our
skip
list
like
a
skip.
These
tests
as
part
of
your
upgrade
procedure
and
those
tests
have
failed
forever.
B
A
Those
four
tests-
yes,
the
ones
down
below
the
seg
apps,
a
full
set,
upgrade
and
master
upgrade
and
HP
upgrade
and
service
upgrade.
Those
four
tests
have
failed
for
ever
like
says
the
dawn
of
of
these
tests,
and
we
tried
last
cycle
to
just
remove
them
as
part
of
the
upgrade
test,
and
we
couldn't
because
there's
some
bug
and
tested
for
us.
So
we
we
just
punted
and
said
this.
These
are
the
acceptable
list
of
failures
that
can
occur
as
part
of
the
upgrade
test.
A
B
A
A
That's
a
tester
for
problem,
not
a
test.
The
testing
for
a
slash
test
problem,
not
a
not
a
cig
coaster,
lifecycle,
ish
problem.
We
have
a
unusual
walk
for
that.
I
am
sure
there
is
one
that
has
existed
for
a
very
long
time
and
have
to
dig
it
up.
But
if
you,
if
you
put
an
actual
item
in
the
meeting,
notes,
I'll
try
and
do
archaeology
to
figure
out
what
the
root
issue
was.
F
A
There
are
some
meta
topics,
because
I
wanted
to
use
this
meeting
for
broader
sig
related
stuff,
which
I
think
is
useful,
like
I
wanted
to
loop
in
other
folks
who
are
still
doing
things
that
are
in
incubation
and
haven't
really
landed.
A
home
right,
I
know,
Minnie
I,
think
mini
cube
is
still
an
incubation.
A
I,
don't
know
anymore
as
well
as
cube
spray
to
try
and
try
to
have
a
landing
home
for
periodic
conversations
that
are
broader
right
so
say,
for
example,
a
good
conversation
pieces
that
I
recently
updated
at
CD
at
the
end
of
this
cycle,
2
3
to
24
upon
recommendations
from
I
I,
don't
want
to
pronounce
his
name
wrong,
but
guy
ho-ho
looked
at
the
last
most
prolific,
maintainer
of
ed
CD.
That
networks,
your
Amazon
right
so
hung
chow
and
Jian
have
moved
on
to
other
places.
A
Now
and
I
knew
that
genome
still
works
on
some
sed
stuff,
but
he
works
I.
Think
at
Alibaba
cloud
now
so
anyways
stating
stuff
like
that
is
useful
for
folks
who
still
have
separate
deployment
models
that
are
outside
of
the
scope
of
inside
of
kaykai
repo,
so
I
updated.
All
of
cluster
Robert
and
I
also
updated
qu
a
DM
as
part
of
this
I
talk
with
Joe
and
pivots
to
make
sure
that
he
was
aware
in
went
Tec
too
as
well.
A
So
they
were
both
aware
so
stuff
like
that
is,
is
useful
to
bring
back
into
the
fold,
and
ideally
we
had
this
outreach.
Originally
in
the
beginning,
we
wanted
to
have
folks
try
to
leverage
Covidien,
that's
inside
of
their
other
installers,
because
we
thought
that
would
be
spread
the
gospel
far
and
wide
so
that
we
don't
constantly
have
multiple
installers
recreating
the
same
sets
of
problems
right
over
time.
C
Think
I
can't
tell
you're
getting
out.
Are
you
trying
to
figure
out?
Should
we
get
things
like
Q's
right,
mini
cube
out
of
incue
bation
and
have
those
be
sort
of
cig
level
projects?
Are
you
trying
to
say
if,
like
how
do
we
consolidate
those
into
the
existing
efforts
under
the
sink,
like
I
can't
quite
figure
out
what
you're
getting
at
here
both.
A
There's
other
there's
other
things
that
are
ongoing
with
inside
of
tested
fur
that
I
think
might
land
a
home
as
a
sub-project.
As
other
folks
know,
we
accepted
the
Moran
test,
dint
cluster
stuff
as
a
sub
project,
and
it
lives
underneath
carbonated,
cigs
org
and
it's
highly
useful.
There's
a
lot
of
people
who
are
very
active
and
prolific
on
it.
There's
always
been
one
core
problem,
and
that
is
that
it's
it's
a
ton
of
bash
and
that
people
really
wanted
to
rewrite
it
and
go
and
bend
the
other
from
Testim.
A
Freh
has
been
putting
a
lot
of
effort
into
what
he's
called
kind,
and
the
kind
is
underneath
of
testing
for
repo
and
I
think
having
general
awareness
of
this
ongoing
effort
and
to
try-
and
you
know,
get
that
into
a
PR
blocking
job
he's
highly
beneficial
to
the
sig.
Because
then
every
pull
request
would
go
through
a
kabillion
set
of
automated
tests.
It
would
basically
be
deployments
of
Covidien
in
this
d-ind
environment,
I.
B
Have
a
couple
of
comments
like
I
agree
about
kind?
Then
we
should
probably
post
it.
It
makes
perfect
sense
it.
The
other
projects,
I
I,
don't
I,
don't
have
historical
backgrounds
on
them,
but
I
do
see
the
same
problem
in
the
project
as
a
whole.
A
lot
of
features
get
to
an
alpha
state
and
then
they
don't
move
forward
like
we
are.
The
only
solution,
like
you
say,
Tim,
is
to
bring
the
people
discuss
it
and
say
like.
Is
it
going
outside
of
incubation
or
like?
C
So
cube
straight
is
an
interesting
example,
because
the
cubes
right
folks
actually
came
to
the
city
and
said
we
want
to
get
promoted
from
incubation
to
a
top-level
repo,
and
this
was
right
when
the
sort
of
steering
committee
was
being
formed
and
sort
of
a
hold
was
put
on.
All
promotions
from
incubation
I
was
also
right
after
cops
had
been
promoted
and
I.
C
Think
those
people
felt
a
little
bit
burned
by
the
process
of
a
different
project
was
sort
of
given
the
opportunity
of
promoted
and
they
were
not
given
that
opportunity,
but
now
that
we
have
a
way
for
sort
of
SIG's
to
themselves
sponsor
project,
and
we
have
the
committee
SIG's
org
I,
think
that
would
be
a
great
home
for
cube
spray.
Assuming
that
those
folks
they
are
still
actively
maintaining
the
project
and
keeping
it
up
to
date
and
be,
you
know,
would
be
willing
to
join
in
this
meeting
and
I.
C
A
So
like
with
the
updated
instructions,
for
example,
for
high
availability
like
we
don't
want
to
have
a
fractal
pattern
of
support
right
where
a
person
goes
to
quest
and
the
installation
is
wildly
different
than
if
they
use
Thule
versus
to
a
B
or
not
even
just
the
tool
fragmentation,
but
the
process
fragmentation
right.
Ideally
we
kind
of
unify
you
know.
Sometimes
some
of
these
things
get
really
highly
opinionated,
but
I
think.
As
long
as
we
have
a
set
of
requirements,
those
opinions
can
kind
of
go
away
right,
we'll
we'll
well
rest
on
the
requirements.
A
There's
the
course
of
things
that
we
can
agree
upon
and
provided
that
it
holds
true.
Then
we
should
be
able
to.
You
know,
rally
folks
around
a
set
of
docs
and
processes
and
tools,
and
then
we
decrease
the
overall
state
space
of
weirdness.
That
kind
of
occurs
in
the
wild
right,
because,
like
cops,
does
things
wildly
different
than
goop
spray
does
and
cube
spray
has
a
cube,
ATM
type
of
deployer,
but
it
doesn't
have
the
high
availability
version
of
it
right
so
I
mean
trying
to
unify
these
things
over
time.
C
C
With
the
you
know
the
we
used
to
have
a
sort
of
side
car
next
to
the
cubelet
to
do
the
certificate
negotiation,
and
we
put
that
into
the
cubelet
right
and
so
having
that
in
the
cubelet
and
be
part
of
this
or
the
standard.
Kubernetes
distribution
makes
that
behavior
more
consistent,
right
and
so
I
think
that
there
are
places
like
that
where
we
can
actually
push
things
into
the
core
components
themselves
to
standardize
things
things
like
component
configuration.
C
If
we
push
that
on
top
of
the
core
components,
then
all
the
tools
get
to
component
configuration
and
that
manner
of
interaction
with
tools
becomes
consistent
versus
how
do
I
figure
out
how
to
get
cops
to
set
this
flag
on
the
cluster
on
the
API
server?
How
do
I
figure
out
how
to
get
cube
admin
to
set
this
slide?
On
the
API
server
right,
so
I
think
those
are
the
things
we
need
to
keep
pushing
on
yeah
and
I.
A
Think
maybe
a
little
bit
of
outreach
would
help,
especially
I
think
right
now,
we're
too
far
into
the
one
12
cycle
for
folks
to
try
and
do
other
things
we're
heading
into
the
home
stretch.
Usually
during
this
end
of
cycle
period,
there's
a
lot
of
frantic
things
that
come
up,
but
in
the
beginning
of
113
I,
think
that's
a
good
opportunity
and
also
leads
into
all
of
the
conference's
at
the
end
of
the
year
right.
A
Where
we'll
be
able
to
like,
as
we
do,
sig
updates,
we
can
give
a
more
broad
update
of
the
globalized
effort,
as
well
as
like
try
to
loop
folks
back
into
the
fold
to
feel
empowered.
I
know
that
there
was
some
hurt
feelings
with
some
of
the
promotions
and
stuff,
but
some
of
the
other
projects
that
even
been
promoted.
B
A
B
B
So
there
is
an
issue
already
but
I
this.
There
was
some
discussion
like.
Should
we
release
cube
DNS
without
manifest
lists
and
I
say
that
that's
a
bad
idea,
because
if
somebody
falls
back
from
coordinates
to
keep
DNS
on
a
multi-artist
coaster,
this
is
going
to
break
for
him
and,
like
I,
wanted
your
opinion
that
should
be
booked
on
this.
No.
A
I,
don't
think
that's
black
already,
the
there
is
we've,
we've
explicitly
stated,
we've
implicitly
stated
I,
don't
think
we've
actually
written
it
down
somewhere
that
we
don't
support.
Multi
large
clusters
like
if
you
were
to
support
a
mixed
mode,
I
mean
you
can
always
do
it,
but
it's
not
a
first-class
type
of
thing
that
we
recommend
or
support,
because
there's
a
lot
of
jiggery
that's
required
in
order
for
you
to
get
that
to
work.
A
You're
basically,
recreating
branded
is
what
you're
doing
and
there's
a
bunch
of
other
semantics
that
come
along
with
it
that
are
non-trivial
to
support,
like
your
images
have
to
make
sure
that
they
land
on
the
specific
ocean.
You
can
do
labeling
properly
for
matching.
So
there's
all
kinds
of
other
things
that
come
along
with
it
that
are
totally
undocumented.
So
yeah.
B
C
B
C
I
suckle
on
there
about
the
the
zoom
meetings,
so
Paris
has
been
bugging
me
about
the
fact
that
our
sig
is
not
following
the
new
procedure
for
house.
New
meetings
are
supposed
to
work.
There's.
This
is
gonna,
be
a
new
zoom
URL
that
we're
going
to
use
which
enables
better
moderation
features.
This
is
supposed
to
be
something
that
the
curates
community
is
applying
consistently
across
the
board.
C
If
the
last
email
from
her
said
that
only
like
two
or
three
cigs
were
actually
doing
it,
but
everyone
should
be
doing
it
so
I'm
gonna
go
ahead
and
update.
The
calendar
invites
going
forward
with
the
new
link
that
I
am
the
process
of
trying
to
generate
with
the
new
account
that
you
just
created
for
for
me,
and
Tim
I
will
share
the
credentials
for
that
with
you.
We
need
to
figure
out
how
to
share
that
across
the
rest
of
the
folks
and
if
we
I
think
we
can
create
multiple
URLs.
C
F
C
A
little
testing,
but
we
have
our
own
account
and
we
would
be
using
that
going
forward
instead
of
the
cluster
office,
account
once
I
finish
getting
it
set
up
so
since
this
week
meetings
every
other
week,
definitely
by
the
time
this
next
week.
Next
many
of
this
happens,
but
hopefully
by
tomorrow
when
the
Wednesday
meetings
happen,
we'll
have
something:
okay,.
A
That'd
be
great
also,
we
should
probably
make
sure
that
to
give
some
sun
set
period,
because
the
way
a
lot
of
people
have
their
reminders
set
up
is
their
reminders
are
copied
from
the
community
calendar.
So,
even
if
you
update
the
invite
that
doesn't
necessarily
get
reflected
onto
their
calendars,
so
we
should
probably
have
like
an
overlapping
two-week
period
or
something
like
that
to
let
people
know
so.
A
A
C
Yeah
once
I
get
it
set
up,
I
can
send
that
out.
I
think
we
are
not
supposed
to
publish
the
link
to
the
meeting
publicly
anymore,
so
it
will
sort
of
it'll
be
in
the
calendar,
invites
for
people
and
we
can
stick
it
into
slack
right.
But
if
you
look
at
like
the
the
sig
pages
and
github,
they
no
longer
have
a
link
to
the
meetings.
Yeah.
A
A
Element
there
is
a
list,
yeah
I.
Think
mailing
list
is
fine,
I
think
that's
partially!
That's
the
main
way
people
will
get
the
information,
so
I
think
that's
completely
reasonable.
I
think
they
just
want
straight
up
web
scrapers
for
other
things.
I
know
canonically
that
the
the
reason
why
issues
evolved
was
from
social
media.
It
wasn't
from
from
scraping
I,
think
they're
doing
extra
precautious.
So.
A
C
Think
so
I
think,
as
I
was
saying,
I
think
we
can
send
it
out
to
me
unless
I
think
we
can
stick
an
agenda.
I
think
we
can.
Secondly,
calendar
invites
I
think
they
just
don't
want
us
to
stick
it
on
github,
where
it's
web
facing
or
posting
on
social
media,
and
if
that
happens,
I
think
we
were
supposed
to
rotate
the
the
link.
So
we
want
to
do
that
as
little
as
possible
as
we're
about
to
find
out
when
we
rotate
it
from
this
one
to
the
new
one.