►
From YouTube: Kubernetes Sig Docs 20181106
Description
Meeting notes: https://docs.google.com/document/d/1Ds87eRiNZeXwRBEbFr6Z7ukjbTow5RQcNZLaSvWWQsE/
The Kubernetes special interest group for documentation (SIG Docs) meets weekly to discuss improving Kubernetes documentation. This video is the meeting for 06 November 2018.
https://github.com/kubernetes/website
--
Chat transcript:
00:13:23 Sil SM: cant we add it to the bot, so that it is checked by automation?
00:13:50 Sil SM: of course buikding localy should always be done
00:14:07 Jennifer Rondeau: The netlify preview builds automatically — we’re talking about the case where it fails. Are you suggesting that we block merge if no netlify preview present?
00:15:22 Sil SM: 👍
00:22:10 Jennifer Rondeau: Andrew, added a comment about the contributing section to your proposal doc
00:22:28 chenopis (sig-docs): Thanks, Jennifer!
00:28:15 Jennifer Rondeau: +1 to Andrew’s proposed approach
00:32:45 Sil SM: +1 too. before I thought there would be a "modeling doc" , having a "read more/details/..." section in the relevant area will be super helpful
00:43:26 Tom van Waardhuizen: I’ll be there
00:43:52 Brad Topol: So are our events part of the contributor summit? Do I register for that?
00:47:34 Sil SM: (jealous about this event. hope to see you in Europe next year )
00:47:45 Brad Topol: I just put myself on the waitlist. Thanks
00:47:57 Kubernetes Sig Docs: Hope to see you next year too, Sil!
00:51:28 Jennifer Rondeau: Brad, Jorge is taking care of you :-)
00:52:06 Jennifer Rondeau: Dominik, I wanna read! I wanna read!
00:52:41 Brad Topol: Thanks!!!
B
A
A
C
A
A
But
he's
doing
good
work,
so
yes,
one
reminder:
please
test
changes
for
poll
requests
locally,
even
if
they
appear
trivial
I
will
raise
my
hand
and
confess
that
I
am
the
reason
for
this
reminder.
I
broke
the
build
rather
spectacularly
last
week
with
a
trivial
seeming
PR.
So
please
make
sure
to
test
in
your
local
environment.
If
you
have
any
question
about
how
to
do
that,
I
think
that
our
Doc's
are
contributor
guide.
Does
a
pretty
good
job
of
telling
how
to
do
that.
A
D
A
I
broke
something
that
that
net
laugh
I,
it
broke,
something
that
wouldn't
have
appeared
in
the
preview.
Build
I
was
working
on
a
process.
That's
transparent
to
users
was
adding
an
owner's
file
to
the
particular
content
directory
and
if
I
threw
an
error,
I
thought
I
knew
why
Irie
committed
and
it's
ended
up
net
laughs.
I,
thankfully,
is
really
strict
with
ownership
parsing.
It
will
not
build
the
site
if
it
doesn't
fully
understand
any
ownership
commands
given
to
it.
D
Of
PR,
not
your
typical,
because
your
typical
PR,
that's
a
I'm,
adding
a
page
editing
a
page.
What
have
you
you
can
go,
look
at
the
preview
build
and
not
feel
like.
You
have
to
run
it
locally.
I
was
just
trying
to
disturb
like
this
was
more
of
an
infrastructure
type
scenario
where
you
really
need
to
know
double-check.
It
I
was
just
trying
to
be
a
little
bit
crisp
there,
because
I
think
there's
at
least
two
different
characteristics.
D
E
Was
one
other
instance
that
happened
recently
I
think
Alex
was
testing
out
switching
from
like
a
PNG
SVG
files
in
the
case
studies
or
something
so
she
removed
the
file
and
then
it
through
an
error
in
the
build.
So
then
I
guess
I
got
merged
accidentally
or
whatever,
but
then
on
that
caused
a
build
problems
down
the
line.
But
the
live
preview
would
have
been
built
right.
D
A
So
I
guess
one
important
note
about
that
process
is
that
net
LeFay
is
not
a
blocker
to
the
github
process.
If
nella
five
throws
errors,
it's
not
fully
integrated
as
a
CI
CD
gate
check
for
our
PR
process
and
I
had
relied
mistakenly
on
the
idea
that
it
was
so
even
if
metla
Phi
is
throwing
errors.
If
you,
if
one
of
us
were
to
approve
a
pull
request
merging
even
if
Matloff
I
is
saying
no,
no
don't
do
it
for
the
love
of
God.
No,
it
will
still
merge.
A
So
that
was
that
was
the
piece
that
was
missing.
For
me
is
Detlef.
I
is
not
a
gay
check
on
our
process.
It
is
possible
to
to
merge
things
that
net
liffe
I
does
not
like
at
all.
So
that's
why
the
reminder
test
locally
or
build
check
the
preview
build
if
the
preview
build.
If
a
preview
build
isn't
available,
it's
a
sign
that
Matloff
is
not
happy
so
make
sure
that
make
sure
that
metla
phi
is
happy
before
merging
appear.
E
A
B
So
many
people
know
to
double
check
to
make
sure
that
the
network
I
preview
is
present
and
then,
if
it's
not
inviting
them
to
ping
on
slack
or
you
know
otherwise,
sort
of
wait
for
help,
and
we,
you
know
maintainer,
so
folks,
it's
been
around
for
longer,
can
step
in
and
try
to
see.
What's
going
wrong.
F
A
F
A
A
A
B
A
B
Prowl,
oh
no,
it's
not
no!
It's
already
in
there.
We
just
need
to
get
those
hooks
back.
I
am
Pierre
Wrangler
next
week,
I
believe
and
and
I
don't
want.
I
could
try
to
get
to
this
this
week,
but
I
don't
want
to
promise
I
think
I
can
promise
by
and
by
next
week
only
because
things
are
as
we,
some
of
us
were
already
discussing
a
bit
chaotic
could
help
EO
and
sometimes
I
get
caught
up
in
the
chaos,
even
when
I'd
rather
be
just
solving
technical.
E
B
G
B
A
A
E
E
E
You
know
these
sort
of
sections
where
you
know
it's
a
certain
area
and
then
it
has
some
quick
links
to
it
to
you
know
the
important
docs
for
that
subject
and
then
you
know
a
call-to-action
button.
So
similarly,
we
were
thinking
of
redoing
the
docs
homepage.
That
way,
because
then
the
the
issue,
what
we
learned
from
the
user
journeys
right
was
that
there
was
too
many
clicks
involved
for
people
to
get
to
the
content
they
wanted
the
links.
E
E
That's
like
you
know
the
production
related
stuff,
like
the
tasks
and
the
troubleshooting
and
the
reference
you
know.
So
this
is
kind
of
what
we've
just
kind
of
sketched
out
so
feel
free
to.
You
know,
give
your
feedback
and
suggestions,
so
this
is
more
I
was
hoping
to
get
kind
of
a
like
a
rough
working
version
up
in
in
the
PR.
So
we
can
like
take
a
preview
of
it
and
then
just
be
able
to
talk
about
it
then
and
then
decide
whether
we
want
it
or
not.
A
That
looks
awesome,
yeah
I
have
some
I
have
some
thoughts,
but
I
will
so
just
to
confirm
the
best
way
to
get
you
feedback
about.
This
is
to
do
you
want
comments
in
the
in
the
Google
Doc.
Do
you
want
comments
in
the
PR?
Where
do
you,
where
it's
going
to
be
the
most
helpful
mechanism
for
feedback
for
you.
E
So
if
it's
about
the
structure
like
from
you
know
how
we're
grouping
things
I
stuff
put
it
in
the
Google
Doc,
because
I
think
that's
where
mostly
shape
I'm
shaping
right
now
and
then
once
once,
that's
decided
then
I'd,
like
any
technical
things
about
the
actual.
You
know,
PR
the
rendering
that
sort
of
thing
that
can
go
in
into
the
PR.
E
A
A
A
So
it
looks
like
there
might
be
an
actual
hugo
bug
involved
with
that.
I
have
I'm
following
up
with
hugo
in
their
project,
so
if
you're
interested
in
following
along
with
the
details,
but
the
takeaway
for
that
is,
if
you
are
working
on
internationalization
or
translating
internationalization
strings,
you'll
have
to
go
through
the
the
regular
english
approvers
process
for
for
final
approval
and
merging
event.
A
A
Just
a
quick
note
now
that
she
is
signed
her
contract.
I
couldn't
say
that
the
CNC
f
we've
hired
a
shabby
de
syndra
to
be
a
contractor
for
us.
Shabih
came
to
our
ducks
print
day
at
right,
the
docks
in
Cincinnati
and
contributed
very
actively
to
the
Spring
Fair.
That
was
how
I
think
Andrew
and
I
met
her
and
that
turned
into
a
conversation
and
we've
been
able
to
hire
her.
B
A
A
E
So
this
week
we're
not
gonna
do
fun
with
modeling,
but
we
wanted
to
talk
about
sort
of
you
know
who
we
think
our
audience
is
because
last
week,
at
the
cig,
Doc's
meeting
there
is
I
think
some
discussion
around
like.
So
if
we
did
fold
this
into
the
community,
IO
website,
like
you
know,
should
we
do
that
or
like
is
this
too
advanced
like?
Is
this?
E
You
know
too
complicated
or
something
like
that
so,
but
that
kind
of,
in
my
mind,
begged
the
question
of
like
who
do
we
think
our
audiences,
because
my
my
assumption
has
been
this
is
not
for,
like
just
you
know,
a
new
computer
science
grad
right
like
like
people
who
use
kubernetes
like
their
company,
needs
to
scale
the
operations.
It's
not
just
gonna,
be
some
Schmo.
That's
trying
to
put
up
a
website.
You
know
cuz.
This
is
that's
way
too
much
overhead
for
that.
E
So
my
my
assumption
has
been
that
they
have
some
sort
of
you
know:
computer
science
or
engineering
or
math
background
or
whatever
and
they've
been
doing
this,
and
this
is
like
you
know-
humanizes
advanced
tool
that
they're
using
for
their
organization.
So
that's
why,
in
my
mind,
my
assumption
was
like
this
would
be
helpful
to
people,
but
then
it
seemed
that
people
other
folks
did
not
necessarily
agree
with
that,
or
there
was
some
other
thing
that
they
were
thinking
that
I
wasn't
aware
of
so
I
Dominic
and
I.
E
Okay
should
be
this
one,
okay,
so
like,
if
you
look,
you
know
on
reddit
or
other
areas
right
and
people
often
have
questions
like
this
one,
for
example,
that
Dominic
found.
Is
there
a
way
to
specify
order
of
containers
to
execute?
You
know
in
a
pod
and
so
obviously
there's
any
containers
right
and
they
can
execute
a
sequence,
but
they
can't
all
end
up
running
concurrently
right.
E
So
so
that's
not
what
you'd
want
to
use,
and
so
he's
you
know
asking
like:
can
you
run
regular
containers
in
a
pod
like
in
start
them
up
in
sequence
and
in
the
general
answer?
Is
no
but
like
he
doesn't
know
this,
because
this
isn't
really
well
documented.
You
know
in
our
documentation,
so
I
feel
like
these
sort
of
questions
should
be
able
to
be
answered
by
just
you
know,
reading
through
our
Docs,
and
they
currently
aren't
another
example,
which
I
think
we
we
showed
before
is
on
the
kubernetes
kubernetes.
E
You
know
issues
right,
so
they
they
asked
this
question
and
Janet
had
to
answer
it
by
pointing
him
to
specific
lines
of
code
in
in
you
know,
in
the
source
code
and-
and
this
was
you
know
like
he
was
not
able
to
reason
about.
Like
the
you
know,
the
status
of
pods
after
you
know
in
a
particular
situation,
so
I
think
our
intent
was
to
help
to
add
some
documentation
that
could
answer
these
questions.
E
So
people
could
figure
this
out,
but
then
there
seemed
to
be
some
pushback
on
I,
don't
know
like
the
diagrams
or
like
how
we're
describing
some
of
these
things,
whether
that's
too
detailed
or
not
and
I
mean
I.
Think
I'd
rather
have
it
in
the
documentation,
because
people
don't
have
to
read
it,
but
it's
there
if
they
need
it.
You
know
and
to
me
it's
just
a
more
matter
of
structuring
the
content.
E
So
you
know,
maybe
we
have
just
like
an
overview
section
that
doesn't
go
in
all
the
details
but
gives
you
you
know
a
clear
idea
of
like
how
things
work
together
and
then
the
standalone
topics
like
Padua
could
then
go
into.
You
know
detail,
you
know
all
those
specific.
You
know
behavior
and
properties
of
pods.
D
This
is
the
this
is
the
the
goal
in
the
modeling
section.
We
think
it'll
help
with
X,
Y
and
C
and
before
you
get
reading
into
it.
Well,
we
kind
of
gave
you
a
heads
up
for
what
this
material
is
gonna
be
for
and
why
it
may
help
or
if
you
feel
like
oh
I,
should
not
even
bother
that
that's
kind
of
what
how
I
interpreted
what'd,
you
just
say
well,.
E
E
I,
wouldn't
even
put
it
in
a
section-
that's
like
separate
from
the
rest
of
documentation
per
se
right,
but
in
like
the
conceptual
section
of
the
site,
we
would
have
an
area,
that's
specifically
overview
right.
So
we
can
talk
about
a
high-level
about
the
architecture
and
what
pods
are
and
this
and
that
so
they
get
a
general
idea
and
then
on
the
standalone
topic
for
pod,
for
example.
E
E
D
More,
the
sprinkled
in
approach,
so
there'll
be
certain
parts
where
so
you'll
take
certain
portions
on
the
modeling
and
then
sprinkle
them
into
the
main
flow,
which
I
think
also
is
fine,
because
someone
can
look
at
it
and
they
can
either
find
it
helpful
or
not,
but
I
don't
think
it
causes
them
pain
that
make
sense.
Yeah.
E
D
E
D
F
I
agree
with
this
idea
that
we
want
to
serve
both
the
beginner
and
the
advanced
users,
so
I
think
this
material
that
you're
working
on
in
the
modeling
group
I
think
it's
appropriate
to
have
some
pages
that
are
for
the
beginner
and
and
then
in
the
pod
topic.
It
can
continue
into
the
advanced
view.
Dominic
here's
a
question
for
you,
so
it
seems
to
me
that
that
often
the
way
you
get
your
insights
is
you
look
at
the
kubernetes
source
code
itself
and
see
how
things
work
and
I
wonder?
F
H
H
Generally,
a
hard
task
for
modeling.
What
do
you
actually
want
to
model
right?
What
lever
do
you
pick
your
abstractions
to
be,
and,
for
example,
if
you
have
a
message
queue
in
a
system,
you
can
have
a
very
high-level
model
that
just
states
that
messages
will
be
delivered
at
least
once
or
you
can
go
deep
down
the
obstruction
stack
and
specify.
H
What
is
the
mechanics
of
the
message,
queue
that
actually
determines,
or
that
actually
results
in
a
message
being
delivered
at
least
once
so
that
very
much
depends
on
on
the
message
you
want
to
come
across,
but
in
general,
when
it
comes
to
kubernetes,
I
try
to
model
the
observable
behavior.
What
are
the
things
that
users
may
see
right?
H
What
is
the
behavior
kubernetes
may
expose
towards
the
user
and
the
general
idea,
and
not
just
to
this
question,
but
the
general
idea
how
I
approach
modeling
is
so
to
give
the
user
a
possibility
to
reason
about
the
behavior
of
the
system.
Simple
example:
if
you
have
a
replica
set
right
with
three
replicas,
for
example,
and
do
you
think
you
are
safe
from
one
node
going
down?
You
have
a
deployment
and
you
specify
rolling
update.
Do
you
think
you
are
providing
a
zero
downtime
right
and
there
are
some
intricacies
in
kubernetes?
H
That
actually
is
says
no
to
both
of
these
questions,
no
replica
set
will
not
guarantee
you
that
if
a
node
goes
down-
and
you
have
three
replicas
that
you're
still
on
because
the
replica
set
doesn't
guarantee
you
that
the
pods
will
be
scheduled
on
different
nodes,
the
same
as
the
deployment.
If
the
deployment
starts
a
rolling
update-
and
you
do
not
have
a
lightness
probe
on
the
container,
the
traffic
will
be
routed
to
this
pod
immediately
after
it
is
ready
which
mean
it
is
up,
but
your
application
may
actually
not
be
ready
right.
H
There
are
a
lot
of
intricacies
to
how
the
individual
pieces
fit
together
and
I
would
like
to
provide
clear
mental
models
that
help
you
reason
about
the
system,
like
the
question
Andrew
showed
on
reddit
right,
so
the
gentleman
had
a
problem
in
mind
method
to
a
solution.
If
the
solution
he
mapped
it
to
was
not
directly
method
to
kubernetes,
so
he
didn't
specify
the
problem
in
detail.
H
And
when
I,
when
I
talk
about
behavior
one
more
example,
which
is
actually
very
interesting
depending
on
how
you
set
up
the
API
server,
whether
it
is
behind
a
load
balance
or
not,
where
the
quorum
read
is
activated
or
not
a
consumer
of
the
API
server
after
creating
an
object,
let's
say:
pod,
the
consumer
creates
or
the
client
creates
a
pod.
The
client
requests
the
pod,
the
quote:
the
pod
is
not
there.
The
client
requests
a
pot
again,
the
pod
is
there.
The
client
requests
a
pot
again,
the
pod
is
not
there.
H
These
are
possible
behaviors,
you
may
see
in
kubernetes,
and
that
you
need
to
guard
your
client
against
that
right.
That
is
something
you
need
to
know,
because
these
edge
cases,
you
will
probably
not
encounter
them
for
quite
awhile,
but
I
would
like
to
provide
models
that
enable
you
to
reason
about
this.
Behavior
expect
this
behavior
and
also
give
you
remedying
or
mitigating
actions.
How
can
I
actually
protect
myself
from
this
behavior.
F
H
E
Another
example,
I
would
say,
is
like
at
CD
right
like
we're
talking
to
an
engineer
before
and
he's
like.
Oh
that's,
just
an
implementation
detail
don't
look
at
that,
but
what
we
found
is
that
it's
sort
of
a
leaky
abstraction
that,
especially
with
high
availability
and
things
like
that,
you
need
to
understand
what
Exedy
is
actually
doing
and
how
it's
you
know
organized.
So
you
know
there
is
some
gray
or
in
terms
of
like
what
people
think
are
just
implementation
versus.
A
So
what
I'm
hearing
from
folks
and
reading
in
chat
is
that
Andrew
I
think
there's
broad
support
and
broad
consensus
for
the
approach
that
you're
proposing
I
I
think
that
the
thinking
back
to
the
push
back
last
week,
I
think
was
trying
to
reconcile
the
work
as
it's
happening,
with
some
of
our
initial
conversations
about
what
modeling
could
do
and
so
trying
to
close
the
gap
between
the
theoretical
and
the
actual
I
think.
But
you
know,
I
agree,
I!
A
A
A
E
A
A
So
here's
what
here's,
what
it
got
so
far
it
puts
in
the
agenda,
there's
a
draft
schedule
from
signature
of
X,
so
we
are
currently
officially
slated
for
two
hours,
but
I
talked
with
Paris
last
week
and
again
this
morning
we
can
request
more
time
requesting
for
hours
seems
like
that
is
a
reasonable
amount
of
time
to
try
and
get
quarterly
planning
done
in
and
sort
of
get
all
the
business
of
our
Sigma
non
a
quarterly
basis.
If
we
can't
do
it
more
than
if
we
need
more
than
four
hours
to
do.
A
B
A
That
was
going
to
be
my
other
question.
Is
that
trying
to
reconcile
the
various
discussions
about
what
we're
doing
in
Seattle?
It
seems
to
me,
like
we
have
two
or
three
different
plans
for
what
we
are
doing
with
our
space
and
our
time
we're
talking
about
doing
a
dock
sprint,
which
I
don't
think
that
we're
gonna
do
I.
Think
we're
gonna
focus
on
quarterly
planning
right
if
I'm,
remembering
last
week's
discussion
correctly.
A
A
A
A
D
A
Let
me
let
me
clarify
that,
because
I'm
I'm
not
sure
about
that
so
I
guess
from
a
high-level
point
of
view.
Thank
you
for
reminding
of
that,
though
Brad
there
is
the
the
unofficial
meet
and
greet
on
Sunday
evening
from
5:00
to
8:00
at
the
garage
in
Seattle.
Just
quick
show
of
hands
are
and
how
many
folks
are
planning
to
go
to
the
Sunday
evening
unofficial.
You
can
greet
okay
well,.
A
Okay,
so
I
I
think
the
very
nature
of
an
informal
meet
and
greet
means.
We
don't
have
to
plan
to
do
anything.
We
just
show
up
and
talk
to
folks
so
for
the
quarterly
planning
Paris
asked
if
we
wanted
to
do
how
we
wanted
to
a
lot
hour
for
hours,
whether
we
wanted
to
do
two
hours
in
the
morning,
two
in
the
afternoon
and
I
said
yeah.
That
sounds
great
with
the
idea
that
the
morning
is
more
of
a
lecture
format
and
the
afternoon
is
more
participation.
A
B
A
Blessing
I,
thank
you,
I
think
that
sounds
like
an
excellent
thing
to
do
during
like
sig
intros
or
like
the
Monday
afternoon.
Meet-And-Greet
in
hello
and
Brad
are
events
part
of
the
contributor
summit.
Yes,
they
are
and
do
you
register
for
that?
That's
a
really
good
question
for
Paris
I
know
that
in
the
past
week,
yes,.
B
G
F
B
F
B
We
can
find
a
way
through
through
you
know,
sick
country
of
X
or
yeah
yeah,
just
to
promote
that
on.
If
people
are
gonna
be
at
UConn
and
would
like
some
hands-on
guidance
getting
started
from
members
of
cig
Knox,
maybe
to
move
it
that
way
and
then
just
in
our
back
pockets,
maybe
have
a
handful
of
issues,
but
not
do
anything
like
coming
up
with
spreadsheets
and
tracking
and
sort
of
the
more
more
there's
another
full-on
sprint
and
make
it
clear
that
it
would
be.
You
know,
sort
of
less
formal.
F
Here's
what
I
think
we
need
I
think
we
need
a
video
that
that
takes
people
through
how
to
how
to
fork
the
repository
and
get
your
local
clones
and
I
think
we
could
get
people
through
that
phase
in
a
video
in
you
know,
20
percent
of
the
time
it
would
take
them
to
read
through
our
written
material,
so
I
just
put
that
out
in
cases
anybody
likes.
You
know
likes
to
work
on
videos
and
would
want
to
do
something.
I
think.
B
My
recollection
is
that
it's
less
intuitive
than
it
used
to
be
you've
actually
got
to
push
your
first
PR
and
get
the
concealing
no
no
flag,
and
then
you
get
the
link
that
you
actually
need,
and
then
everything
sort
of
cleans
up
and
showing
that
is
I
would
I'm
guessing
a
lot
easier
for
people
to
get
then
being
told
it,
and
we
had
people
tripping
over
it
at
the
right.
The
docks
in
Cincinnati
I
think
everybody
tripped
over
it,
except
maybe
one
person
and
I
think
that
was
shoddy.
Yeah.
B
B
A
H
There
is
one
thing
so:
Andrew
and
I
are
planning
to
release
another
blog
post
this
week
about
the
API
server.
It's
actually
a
three
part
series
where
we
talk
about
the
API
server
in
general,
then
the
implementation
of
the
kubernetes
api
and
the
implementation
of
the
object
store
on
top
of
etcd
and
the
first
part
of
this
series
will
heavily
rely
on
specifying
the
abstract
behavior
of
the
API
server
without
going
into
any
implementation
details,
we'll
also
heavily
rely
on
formal
specification
language
code
alloy.
H
If
you've
got
the
time-
and
if
you
can
please
do
check
out
the
blog
post,
we
will
send
it
around
and
be
super
critical
about
it.
I
am
very
curious
if
it's
kind
of
an
experiment
and
I'm
very
curious
if
this
high-level
description
of
the
behavior
of
the
IP
API
server,
without
actually
talking
about
anything
in
concrete,
like
resources
versions
and
how
the
data
is
represented,
is
useful
for
anybody
to
help
them
reason
about
the
behavior
of
the
system.
Interesting
yeah,.