►
From YouTube: 20210421 SIG Arch Prod Readiness
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
All
right
welcome.
Everybody
today
is
april
21st
2021,
and
this
is
the
kubernetes
production
readiness,
subproject
meeting
and
let's
get
started
so
just
a
couple
things
on
the
agenda
today,
which
I'm
displaying
here
we
had
some
action.
I
had
some
action
items
that
I
hadn't
followed
up.
A
Everybody
else
followed
up
on
theirs,
I
think-
and
so
I
did
thankfully,
and
the
one
that
we
really
need
to
go
over
here
is
review
the
survey
so
a
year
ago,
just
about
a
year
ago,
a
little
more
than
a
year
ago,
we
did
a
survey
as
we
were
forming
this
group
about
different
versions
of
kubernetes
and
whether
people
ever
had
to
roll
them
back
whether
they
enable
things
in
production.
A
A
So
I
have
updated
that
survey
with
a
few
changes
and
we
can
review
it
here.
So
I
think
before
I
I
assume
it's
showing
yeah.
Okay,
so
just
to
say
one
more
thing
is
before
we
do
that
we
also
can
consider
removing
questions
if
we
think
that
they're
not
useful,
shorter
surveys
are,
you
know,
get
better
responses,
but
anyway,
just
something
to
think
about
so
changes
since
the
last
one.
Basically,
I
added
a
little
blurb
here,
oops
just
to
say,
hey,
we
started
this
year
ago.
A
One
question
is,
we
originally
talked
about
doing
this,
like
maybe
twice
a
year
and
and
then
we
didn't,
but
of
course
we
didn't
get
to
where
we
are
mandatory
until
121,
which
is
the
release
that
just
came
out.
So
I
don't
know
that
the
results
would
have
been
all
that
different
over
the
last
year
than
the
first
one.
A
I
mean
we
did
some
reviews,
so
maybe
it
would
have
been
a
little
different,
but
do
we
think
we
want
to
do
this
like
semi-annually?
A
And
you
know
it's
more
work,
it's
not
hard
to
run
the
survey,
but
analyzing
the
results
actually
takes
some
time.
I
have
a
by
the
way
in
for
those
I
know
because
atlanta,
you
weren't
involved
at
the
time,
so
you
probably
never
saw
this.
I
added
a
link
here
to
the
results
from
last
time.
A
So
we
got
about
80
responses
last
time.
Hopefully
we
can
get
better
spread,
and
maybe
I
can
work
with
bob
killen
at
the
in
controversy
to
try
and
see
if
he
can
has
ways
to
communicate
it
better
or
something,
but
anyway
yeah
this.
This
is
this
is
shared
now
with
kdep,
so
you
should
have
access
to
it.
But
anyway
you
can
look
through
it.
If
you
want,
but
anyway
we
don't
need
to
do
it
now.
B
Yeah,
I
know
I've
talked
to
bob
about
this
before,
because
instrumentation
was
trying
to
run
a
survey
at
some
point,
and
my
understanding
is
that
I
think
google
forms
aren't
available
in
china
necessarily
so
I
think
they
have
like
some
way
to
send
a
linux
foundation
form
out
to
contributors
who
can't
access.
Google
forms,
but
I
don't
know
what
the
process
is
for
that
and
we'd
have
to
convert
it.
So.
A
A
Then
I
load
into
bigquery
and
lunge
around
with
the
data.
So
oh
kirsten's
here
hi
kirsten
and
hey.
So
you
know
I
I
don't
know
that
I
want
to.
A
I
mean
that
would
be
nice,
but
I
don't
know
what's
involved
with
that,
if
you,
if
you
want
to
check
that
out,
you're
welcome
too
okay
anyway.
So
what
do
you
think
annual
semi-annual.
C
I
think
I'm
liking
annual
right
now.
I
wanna
have
time
for
two
significant
things
to
hit
before
we
give
our
next
one.
I
want
to
see.
I
think
we
got
about
half
the
enhancements
before
we
required
it.
Then
we
got
about
half
getting
preview
and
119
to
120.,
but
I
want
to
before
we
do
the
next
one.
I
want
to
make
sure
that
clouds
have
have
hopefully
caught
up
to
121
before
we
issue
the
next
one
I
know
like
gk
and
eks.
C
They
tend
to
lag
behind
a
bit
and
I
would.
B
C
Have
the
first
round
of
perma-beta
things
removed
so
122.?
I
would
like
to
see
just
how
big
that
comes
back
right.
It
was
actually
an
action
item
out
of
the
first
time
we
ran
this
survey
to
make
sure
we
got
that
landed
and
make
sure
it
stuck
yeah.
That's
what
I
want.
D
I
think
I
think
I
I
agree
with
david.
I
think
like
running
it
like
semi-annually
will
be
too
frequent
because,
especially
with
like
free
releases
per
per
year,
it
means
like
running
every
one
and
a
half
release,
so
it
seems
too
frequent.
B
Plus
one
for
me
for
doing
it
annually
should
we
have
an
agenda
item
for
what's
going
on
with
the
122
removals?
I
guess
that's
not
really.
Our
group,
I.
C
122
removals
yeah;
okay,
mostly
I'm
just
worried.
C
So
we
tried
kubernetes
dev
kubernetes
users,
the
group
out
on
the
discourse
page.
Oh
it's.
B
C
Well,
so
we
have,
we
have
a
doc
page
that
describes
it
and
describes
the
changes
that
are
required.
It's
come
up
at
api
machinery
meetings.
We
have
broadcast
to
those
mailing
lists.
A
B
Yeah,
I
think
that
would
be
super
helpful
because
it
makes
me
nervous
as
like,
not
really
knowing
what's
in
scope,
what's
not
in
scope.
What's
on
the
plan
for
being
axed?
What's
not
because
I've
been
looking
at
various
things
like
well,
we
don't
want
things
to
be
in
perma-beta
forever,
so
any
feature
flags
that
have
been
sitting
around
in
in
beta.
We
need
to
get
rid
of
this.
C
C
B
C
So
it's
out
there,
it
did
enumerate
the
list,
so
I'm
happy
to
resend
it
and
then
yes,
we
have
a
docs
page
all
about
the
ones
that
are
getting
removed.
Why
they're
getting
removed?
What
defaults
are
changing?
How
you
react?
That's
already
out
there
as
well
and
was
included
in
the
email
so.
A
Okay,
all
right,
let's
keep
on
with
our
little
survey,
then
so
we
all
great
annuals
good.
Okay,
this
hasn't
changed.
A
Here
I
added
cops
and
rancher,
because
those
were
two
that
were
on
that
showed
up
as
little
clusters
in
the
other
category
and
then
I
split
self-hosted
into
on-prem
and
on
a
cloud
provider,
meaning
like
you,
can
run
obviously
open
source
kubernetes
on
any
cloud
provider.
And
I
thought
that
might
be
a
useful
distinction.
But
I
don't
know
what
you
guys
think
yeah.
A
A
A
Nothing
changes
here.
Nothing
changes
here,
versions
used
to
be
1.10
or
earlier
I
bumped
that
to
1.13
or
our
last
survey.
A
I
kind
of
picked
113
earlier
because
in
our
last
survey,
25
almost
25
percent
of
the
people
who
are
on
a
113.
ago.
A
Sorry
too
many
things
here
anyway,
I
was
gonna
fight,
see
if
I
could
find
that,
but
that's
fine
that
was
kind
of
the
cut
off
was
113
and
then
it
you
know
it
went
up
to
117
or
something
there
and
like
anyway.
If
people
think
that's
still
too
old,
we
can.
We
can
bump
it
up
further,
probably
121,
which
is
the
current
release.
One.
B
Thing
that
sorry
jumping
up
a
little
bit,
I
noticed
you
know
we
ask
for
like
number
of
nodes
and
I
think
number
of
clusters
we
don't
ask
for
numbers
of
pods
per
cluster.
I
don't
know
if
that's
relevant
here,
but
I
think,
like
the
density
of
packing
in
a
given
cluster,
also
gives
you
a
picture
of
how
the
cluster
is
being
used.
B
B
A
Okay,
I
see
what
you're
saying
yeah
you're,
not
asking
for
what's
the
density
of
paths
you're
just
saying
how
many
pots-
yes,
I
don't
know,
what
do
other
people
think.
C
I'm
interested
in
how
we
would
use
that
information
with
the
other
questions
we
have,
I'm
certainly
open
to
it,
but
you
know
how
would
be
how
would
we
relate
that
to
how
frequently
you
roll
back,
for
instance,
right
the
next
questions
that
we
have?
I
guess
I
don't
know
if
they're
the
next
questions,
but
there
are
questions
about,
have
you
rolled
back
and
why?
How
would
the
number
of
pods
help
us
analyze
it
differently.
A
So,
let's
look
at
what
we
did
last
time,
so
last
time
we
looked
at,
we
looked
at
what
did
we
slice
by
those
two
by
those
two
by
those
two
dimensions:
oldest
version
versus
newest
version
based
on
the
number
of
nodes
under
management
number
of
clusters
in
our
management?
Some
of
these
have
obvious
kind
of
like
you
know.
If
you
are
got
more
than
10
000,
if
you're
more
than
a
thousand
clusters,
then
your
newest
is
the
latest
one
and
your
oldest
is
the
oldest
one.
A
You
know,
so
even
that
wasn't
super
useful.
What
else
did
we
slice
that
way.
C
What
we
used
the
nodes
for
was
to
try
to
decide
where
we're
going,
to
focus
and-
and
I
remember
at
least
I
had
not
considered
it
when
we
started,
but
when
we
got
into
reviewing
the
results,
it
became
clear
that
it
doesn't
really
matter
what
someone
with
more
than
10
000
nodes
under
management
does
because
they've
done
all
the
things,
if
you
have
more
than
a
thousand
clusters,
you
will
have
had
to
roll
one
of
them
back,
just
it
always
happens,
and
they
appear
to
use
some
of
our
debugging
differently
right,
like
when.
C
I
remember
the
middle
set
of
people
actually
used
events
for
debugging-
I
don't
know
if
you
remember
that
it
was.
It
was
weird
and
noteworthy
that
like
people
actually
looked
at
the
logs
to
figure
out
what
was
going
wrong
and
that
really
surprised
me
at
the
time,
but
we
were
able
to
relate
it
to
the
number
of
nodes
that
they
had
do.
We
think
pods
will
tell
us
something
new
and
distinct.
B
So
I
don't
think
that
they
would
necessarily,
I
think,
it's
a
fair
and
a
reasonable
question.
I
don't
know
they
would
necessarily
give
us
another
dimension
to
slice
on
it's
just
another
piece
of
like
I
mean,
for
example,
ops
versus
dev
versus
devops.
Like
should
we
take
that
out
because
I
don't
think
we're
slicing
on
that,
but
it's
demographic
information
that
we're
currently
collecting
so-
and
I
think
this
is
like
the
only
real
survey
that
we're
doing
where
we're
gathering
this
data.
So
I
mean
I
would
be
totally
fair.
C
Were
trying
to
look
at
ops
versus
devops
to
inform
the
prr
survey,
questions
around
you
know:
do
you
want
to
focus
on
logs
or
do
you
want
to
focus
on
metrics
and
which
sorts
of
things
will
help
which
sorts
of
roles
right?
C
A
B
B
Think
that
the
the
framing
here
is
just
not
right,
because
right
now
we
have
you
know,
for
example,
like
dev
ops,
devops.
B
I
think
what
you're
pointing
at
more
is
perhaps
application
platform
like
you
know,
direct
cluster
administration,
that
kind
of
thing,
and
I
don't
actually
think
that
we
get
that
out
of
dev
versus
ops
versus
devops.
C
So
you're
thinking
get
rid
of
the
three
that
we
have
and
shift
it
to
to
what
application.
B
Application
development
like
operating-
and
this
is
not
very
sound
bitey
but
like
operating
kubernetes
as
a
platform
either
for
you
know,
like
one
team
or
for
others
to
consume
kubernetes
like
there's,
basically
like
a.
A
Yeah,
so
so
what
I
might
suggest
atlanta
like
like
it
sounds
like
you,
have
some
ideas
around
that
you
can.
You
should
have
edit
privileges
on
this
form,
so
you
could
go
and
add,
add
a
second
question
and
then
we
can.
We
can
all
offline
pick,
you
know,
or
I
don't
think
they.
I
don't
think
it
has
comment
ability
like
a
real.
You
know
like
docs,
okay,.
A
Yeah
some
ideas
yeah,
we
don't
need
to
do
that
live,
but
that's
a
great
that's
a
great
point
that
this
might
not
really
capture
what
we're
trying
to
get
out
of
that.
So
david.
You
are
interested
in
how
those
things
dictate
sorry
just
a
second
kirsten
about,
because
we
have
later
questions
about
what
tools
people
use
to
debug,
and
so
you
were.
You
wanted
to
see
how
these
correlate
to
that
and
then
the
other
use
of
that
data
is
like
we
application
developers
who
fill
out.
A
C
Yeah,
that
was
my
only
memory
of
how
we
used
that
demographic
information
before
yeah.
Okay,
sorry-
and
it
was-
was
it
kirsten.
F
Oh
yeah,
I
just
had
a
general
question
of
somebody.
Spying
on
the
meeting
is
the
purpose
of
the
survey
to
sort
of
survey
more
end
users
of
kubernetes
to
inform
what
the
prr
is
going
to
sort
of
ask
for
from
the
the
kep,
the
kep
authors,
to
make
sure
that
you
know
you're
asking
them
the
right
questions
for
the
benefit
of
the
users.
C
A
Okay,
you
know-
I
I
don't
know.
Are
these
really
useful
for
that?
This
was
more.
I
think
our
curiosity
than
our
like,
like
I
will
admit,.
C
B
C
So
it
helped
explain
that
I
don't
think
we
used
to
slice
anything
else,
because
after
we
gathered
it,
we
realized.
None
of
us
wanted
to
allow
this
to
become
some
sort
of
marketing
tool.
C
B
A
C
I'd
be
okay,
snapping
that
question
it
does
explain
some
of
the
version
variants
which,
when
I
manually
looked
at
the
results,
I
found
helpful.
A
A
I
can
look
after
because
that
we've
only
shared
with
us
before
now,
let's
share
with
kdev
yeah.
Okay,
let's
kill
this
question.
I
don't
think
we
need
it.
I
don't
think
we
used
it
for
anything.
A
All
right
versions:
okay,
okay,
so
so
we
don't
like
get
to
the
nitty
gritty
of
like
you
rolled
what
versions?
Did
you
roll
back
in
production?
What
versions
did
you
roll
back
in
staging?
What
versions
did
you
roll
back
like?
We
just
asked
you
to
answer
these
questions,
give
us
the
reasons
and
then
what
version
did
you
roll
back
from?
A
I,
I
think,
that's
okay.
I
don't
know
that
we
need
you
know
we
could
stipulate.
I
I
did
stipulate
in
here.
So
the
changes,
sorry
to
get
back
to
like
changes
from
last
time.
I
I
obviously
changed
the
versions
I
stipulated
in
these
questions
now
for
a
reason
other
than
to
test
how
to
roll
back
and
whether
it
works,
because
that
was
one
of
our
main
others
that
people
added
and
we
don't
really
care
about
that.
C
Do
I
agree,
do
you
agree
with
a
good
change?
I
do
remember
wishing
that
we
knew
when
something
didn't
work
right,
and
that
was
the
reason
that
they
rolled
back.
I
remember
wishing
that
we
knew
what
piece
didn't
work
right.
C
C
That's
fine.
The
only
other
time
I've
used
is
to
like
run
after
school
activity
surveys,
so
yeah.
A
B
A
A
A
A
A
C
A
A
A
I
had
I
had
to
hand
code
these
disabled
the
test
reasons
right.
So
that's
why
you
know,
but
nobody
answered
it,
which
makes
sense
to
me
that
that's
not
something
people
would
do
that
for.
C
Now
I
would
like
to
change
the
question
that
we
had
about.
Are
our
developers
and
organization
allowed
to
rely
on
beta
features?
I'd
like
them
to
give
a
list
of
the
beta
features
that
they're
allowed
to
rely
on?
Oh
nobody's,
going
to
do
that
david.
Come
on
no
one's
going
to
say
that
you
don't
think
so.
A
C
E
B
A
C
A
C
If
they
give
me
a
list
I'll
be
able
to
pick
out
what's
a
feature
and
what's
an
api
reader,
oh.
C
B
C
A
C
A
B
I
think
this
is
this
is
just
a
summary
section,
so
it
looks
like
they
have
the
number
of
responses
per
question.
Hopefully
we
won't
get
emails.
If
you
scroll.
A
B
A
I
I
need
to
do
what
is
this
one
disable
disabled
beta?
Oh
yeah,
these
are
all
percentages.
This
is
like
eight
percent
right.
It's
showing
that
already.
Okay,
here
we
go
disabled.
Ga
in
development
is
yeah
two
and
a
half
percent
disabled
ga
and
staging
disabled
ga
production.
I
just
couldn't
read
my
own
charts.
Thank
you.
So
these
are
percentages
here,
so
they're
up
they're,
pretty
small.
The
biggest
thing
is
disabling
data
in
production
at
7.7
percent.
A
F
A
Okay,
did
we
want
to
ask
anything
about
oops
this?
I
didn't
finish
this
question.
What
do
we
want
to
ask
about
the
data
feature.
A
A
A
A
A
All
right,
and
then
I
updated
this
question
a
little
bit
I
sort
of
in
alphabetical
order,
just
because
I
thought
that
was
more
politic
and
I
added
datadog,
which
was
a
little
cluster.
We
had
people
wanted
to
distinguish
between
fluentd
and
fluent
bit.
So
I
added
fluent
bit.
I
had
a
new
relic.
Those
were
little
clusters
we
had
in
our
freeform
responses.
A
You
know
one
thing
we
never
did
and
I
we
never
like
actually
followed
up
on
that.
We
asked
people
their
email.
We
had
a
good
20
people
give
their
emails,
saying
they'd,
be
willing
to
talk
to
us
for
follow-up,
and
we
never.
I
guess
just
it
was
just
too
intensive
the
amount
of
work
involved,
but
we
never
like
followed
through
on
actually.
C
C
Absolutely
I'm
most
interested
in
whether
the
same
people
respond
tracking
their
response
over
time.
A
C
Yeah,
well,
I
think
this
is
that
document.
I
think
it's
the
you
answered
last
time.
Please
answer
this
time
and
let
us
know
you
know
the
difference
or
we
write
once
you
answer.
A
A
A
B
A
Yes,
okay,
that's
cool
and
kirsten.
Yes,
a
number
of
these
do
allow.
A
Multiple
selections,
the
ones
that
are
square
or
multiple
selections,
ones
that
are
round
are
radial
buttons,
so
yeah
many
of
them
are
multiple.
C
I
think
there
was
one
more
spot
where
I
just
saw
you
scroll
past,
where
you
still
had
the
one
dot
x
and
x,
plus
one
and
oh
yeah.
Okay,
yes,.
A
There
are
probably
but
there's
a
few
places
that
will
have
that,
but
we
can
we.
B
C
A
Cool
then
I
will
go
through
it,
one
more
time,
offline
and
fix
those
things
and
what
then,
last
time
I
sent
it
out
to
the
kubernetes
kdev.
Of
course
I
sent
it
to
the
kubernetes
discuss
thing.
A
I
probably
poked
it
on
slack
in
different
channels
like
k,
users
and
whatever
else
there
is
on
slack,
but
I
I
guess
we
can
do
all
that
and
we
can
talk
to
contributes
about
how
we
might
get
it
out
there
and
widely
shared
and
and
hopefully
we'll
get
even
more
responses.
This.
C
Time
yep,
I
guess
the
only
thing
I
would
add
is
maybe
ask
them
whether
there
is
some
sort
of
normal,
easy
way
to
take
the
email
list
we
have
from
previous
respondents
and
see
if
they'll
take
the
updated
one.
C
That
works
fine
for
me,
okay,
and
we
did
say
that
we
would
use
it
for
purposes
like
that.
So.
A
A
All
right,
any
other
things
we
wanted
to
talk
about
today,
there's
nothing
really
else
on
the
agenda
other
than
you
know.
Somebody
review
this.
I
assigned
it
to
the
three
of
you
on
the
team,
so
I
don't
think
it's
anything
controversial.
It's
pretty
trivial
and
anything
else.
Anybody
else
wants
to
bring
up.
B
I
brought
up,
I
think,
in
the
channel
at
some
point.
How
do
we
do
now
that
things
are
sort
of
flowing
in
for
prr?
How
do
we
divvy
them
up
like,
for
example,
for
a
thing
that
somebody
did
the
prr
review
in
alpha
that
wants
to
graduate
to
beta?
Is
that
going
to
be
the
same
reviewer?
Is
that
going
to
be
a
different
reviewer?
A
A
C
E
A
A
F
F
B
A
Have
they
published
the
schedule
for
the
release,
yet?
No
okay,
so
we
don't
know
when
our
deadline
is
yet.