►
From YouTube: Kubernetes WG LTS 20190307 APAC TZ Meeting
Description
Introductory meeting for APAC tz
WG-LTS Charter: https://github.com/kubernetes/community/blob/master/wg-lts/charter.md
Meeting frequency: biweekly
Mailing List: https://groups.google.com/forum/#!forum/kubernetes-wg-lts
Meeting and Agenda Doc: https://docs.google.com/document/d/1J2CJ-q9WlvCnIVkoEo9tAo19h08kOgUJAS3HxaSMsLA/edit?ts=5bda357d#
A
A
B
C
C
B
B
So
yeah
I
think
for
today
our
agenda
is
just
to
yes.
John
about
is
about
his
thoughts
and
to
just
sort
of
bring
everyone
in
the
core
I
hate
John
up
to
speed.
On
now
on
what
we'd
be
talking
about
in
the
US
time
zone
call
so
yeah.
We
yeah
I,
think
I
think
that
that
probably
is
one
of
the
most
valuable
parts
for
consumers
of
having
a
LCS
is
or
something
like
a
Nokia's
is
yeah,
just
being
able
to
certify
specific
versions.
B
C
C
C
A
So
there
is
a
survey
out
that
is
trying
to
get
this
without
exactly
asking
people
about
solutions.
It's
trying
to
get
the
the
currents.
How
is
the
user
community
using
kubernetes?
What
are
their
problems
right
now
with
respect
to
release,
is
using
upgrades
etc?
So
we
have
a
survey
dachau.
It
should
take
about
10
minutes.
Someone
said
they
finished
it
in
four
minutes.
So,
if
forwarded
to
anyone
who
do
you
think,
there's
two
you
know
would
help
you
know
to
present
your
ideas
or
represent
newer
ideas
of
how
kubernetes
is
being
used
today.
A
So
I
think,
based
on
that
way,
the
group
working
group
will
look
at
the
proposals
to
see
what
are
the
biggest
problems
that
the
solution
that
we
come
out
LTS
just
in
the
name
of
the
working
group.
But
the
idea
is
to
see
how
we
can
change
the
support
and
release
cadence
to
match
what
the
users
are.
Looking
for,
yeah.
B
A
No,
that's
not
make
sense
I
taken
what
the
way
you
spoke.
John
was
completely
clear
to
me
in
a
way
because
I
worked
as
an
ops
for
two
years
on
kubernetes
and
I
had
similar
experiences
of
upgrades,
and
just
you
know,
being
very
optimistically
trying
to
upgrade.
Very
you
know
you
know,
as
in
was
as
I
went
kubernetes
releases,
but
it
takes
me
a
lot
more
time.
Even
after
being
a
contributor
for
five
four
years,
I
took
me
huge
amount
of
pains
to
do
the
upgrades
and
I
was
not
always
successful.
A
I
think
last
time,
I
was
two
or
three
releases
behind
in
being
able
to
upgrade
so
yeah.
That's
the
I
can
totally
understand
what
you're
coming
from
an
please
feel
free
to
share
any
of
your
experiences
you
have
in
the
in
the
survey
and
also
there's
a
freeform
text
field
that
tells
us
you
know
you
want
to
say
something
else
also
which
helps
people
cover
aspects
that
are
not
covered
in
the
questions
and
will
help
us
form
questions
for
the
future
survey.
Is
this
the
hidden.
C
A
B
A
B
A
A
So
from
that
perspective,
even
if
you
meet
you,
your
users
who
use
cook
your
solutions,
maybe
they
can
also
give
you
a
I,
didn't
I,
don't
want
you
to
take
down
kubernetes
ever
and
that's
their
perspective,
that's
a
good
perspective,
and
if
you
have
gone
through
the
survey,
if
you
have
any
feedback,
please
feel
free
to
share
the
feedback
saying
I
get.
This
question
was
very
good
versus
hey.
This
question
wasn't
clear
to
me
at
all:
I
couldn't
understand
what
I
was
supposed
to
answer
or
suggestions
on.
A
B
Yes,
aie
I
also
I
gave
a
talk,
devops
ties
Sydney
couple
weeks
ago,
so
yeah
I'm,
the
obvious
one
view
like
the
main
thing
is
here
that
we
are
still
talking
about
and
sort
of
Tim,
st.
Clair's
sort
of
stable
devel
branch
and
a
proposal,
but
Liuba
me
who
worked
for
vmware
and
he's
one
of
the
maintained
as
a
cube
ADM.
B
He
has
sort
of
had
some
closing
thoughts,
so
he's
gonna
write
up,
I,
suppose
of
his
as
well
too,
and
to
sort
of
indicate
yes
to
give
us
all
something
to
talk
about
basically,
I
think
and
some
of
the
other
discussions
that
are
kind
of
relevant.
Are
the
there's
been
a
couple
of
discussions
about
just
general
stuff
that
needs
to
improve
for
us
to
be
able
to
anything
more
more
than
sort
of
through
the
three
months
they
released
I
was
we
had
now
so
the
things
we
talked
about
are
like
dependency
management
right
now.
B
You
know
if
you,
if
you
keep
up
with
the
release
train,
you
also
need
to
keep
up
with
a
whole
bunch
of
dependencies,
the
dolçaina
D.
You
know
Carter
you're
using
Carter.
You
know
all
that
stuff.
You
know
that
just
sort
of
the
tip
the
version
divinsky
matrix
is
surprisingly
complicated.
Well,
not
surprisingly,
but
it's
quite
complicated
and
it's
yeah
at
the
moment
it's
poorly
communicated.
Oh
sorry,
we
got
something
else.
Hi
Jade,
welcome
to
the
impact
on
zone
WD,
LTS
meeting.
Thank
you
for
coming.
B
A
B
Yeah,
I'm.
I
think
that
we
put
a
there's
a
an
issue
in
cig
release
about
about
dependency
management,
about
updating
the
external
dependencies.
So
I
think
it
is
something
that
will
get
some
focus,
but
it's
a
sort
of
thing
where,
as
the
more
that
we
talk
about
what
we
would
actually
functionally
need
to
be
able
to
make
to
be
able
to
make
longer
lasting
versions
than
three
months,
then
that's
where
we
sort
of
start
going
on
the
actual
actions
that
the
kubernetes
project
can
take
right
now,
one.
A
One
recommendation
would
be
you
guys
to
start
also
contributing
the
EU
version
of
cig
release
to
go
and
raise
this
issue
saying
hey.
What
does
kubernetes
actually
mean?
What?
Where
does
the
project
boundaries
and
to
say
see
the
the
release
of
kubernetes
means
these
things,
and
it
does
not
care
about
a
version
of
version
of
this
version
of
kubernetes
or
that
version
of
kubernetes,
but
this
version
of
what
CD
or
that
version
of
docker.
A
So
where
does
the
boundaries
of
kubernetes
release?
Where
does
the
ownership
and
responsibility
of
the
kubernetes
project
end
and
that's
something
that
CPM
and
cig
release
have
to
decide?
But
as
long
as
the
users
don't
show
their
problems
like,
if
you
are
a
vendor
or
if
you
are
someone
who
is
distributing
a
product
installing
it
on
say?
As
with
the
kubernetes
as
a
platform,
you
are
also
not
just
bundling
Cooper
Nativity
are
also
responsible
for
the
docker
you're,
also
responsible
for
the
underlying
OS.
A
You
know
the
principle
for
the
goal
that
you're
building
this
projects
fit
and
hcd
hcd
and
sort
of
support
before
or
docker
and
sort
of
support,
though
kubernetes
alien
support
here
still
and
they're
still
using
an
on
you're
still
breaking
the
audit
cycle.
So
your
customers
will
run
their
audits
and
file,
find
that
you're
using
an
on
supported
version
of
talker,
and
then
you
get
stuck
because
kubernetes
does
not
support
another
version
of
tucker.
B
Of
the
things
I
think
one
of
the
people
in
Australia
is
the
I
mean
Australia
is
a
small
enough
market
that
almost
everybody
using
communities
are
effectively
buck,
end
users
or
enterprises
that
are
consuming
communities
as
like
operators
doing
things
on
top
of
communities
by
just
running
services.
Rather
than
doing
all
platforms,
I
think
as
a
lysine
we're
a
little
a
normal
thing
that
we
are
building
a
platform
on
top
of
communities,
and
so
some
of
our
use
cases
are
a
little
bit
more
like
a
cloud
provider.
B
A
I
agree,
but
even
if
I
don't
see
the
counter
size
important,
it
start
a
vendor.
Who
is
the
smaller,
smaller
vendors
I
call
them
vendors,
but
they're
I,
don't
know,
what's
the
good
name
for
them,
but
if
you're
having
a
product
lots
of
products
that
get
deployed
as
cells
or
VMs,
or
you
know
that
run
that
depend
on
an
underlying
platform
on
a
rail
or
so
say
or-
and
you
know,
put
your
binary
on
there
and
your
binary
is
your
service
and
your
product
that
you
deploy
to
a
customer
and
in
in
the
future?
A
We
out
here
seeing
a
pattern.
People
might
do
that
on
kubernetes.
They
might
bundle
kubernetes
along
with
the
product
and
give
it
off
to
hospitals
or
give
it
off
to
banks
saying
that
okay,
this
is,
if
you
want
an
audit
software
I
know
the
soft
orbit
software
on
a
kubernetes
platform.
Once
you
do
that,
you
are
tied
to
the
kubernetes
release,
because
your
product
is
now
you're
shipping
kubernetes,
along
with
everything
else.
So
you
are
responsible
for
any
issues
on
that
kubernetes
layer.
A
And
if
there
is,
you
know
if
the
banks
do
internal
audits
and
they
find
that
certain
products
have
CVS
that
are
not
back
ported.
You
are
your
product.
Is
that
yet
you
are
recommending
that
particular
version
of
the
product,
then
you
get
stuck
so
no
matter
what
version
of
kubernetes
or
what,
whichever
type
of
vendor
you
are
small
or
big.
If
you
are
smaller,
you
have
less
capacity
to
address
those
issues.
A
You
don't
have
tens
or
hundreds
of
people
to
work
on.
You
know
container
runtimes
and
operating
system,
patches
and
track
of
CVS
in
a
small
library
that
small
go
package
that
kubernetes
uses,
and
so
there's
a
lot
of
lot
of
things
to
think
about
when
you
are
thinking
about
from
support
and
a
small
vendor
and
then
from
kubernetes
perspective,
cig
release
and
CPM
are
the
best
places
to
raise
the
issue
of
creating
a
fine
boundary
on
where
you
should
where
well,
where
they
should
solve
this
problems.
B
If
you
are
coming
in
and
out
welcome
to
the
WDS
meeting
things
that
you
need
to
know,
the
meeting
is
being
recorded,
conduct
and
Dhamma,
let's
just
paste
at
the
note
that
are
linked
to
the
meeting
minutes
in
there.
If
you
wanna,
hang
around
that's
great,
please
feel
free
to
drop
your
name
in
the
many
minutes
so
that
we
can
know
you
were
here
and
if
you
would
like
to
feel
free
we're
just
talking
at
the
moment,
because
no-one
else
is
not
because
we
are
very
interested
in
anything.
Anyone
else
has
to
say.
A
E
E
B
Yeah
nice
to
meet
you
man,
yes,
sir,
my
name
is
Nick
Young
I
work
at
a
lysine
and
I'm
one
of
the
co-chairs,
along
with
Davao
and
Tim
pepper
and
Queen
OH,
so
yeah.
Then
we
and
the
point
of
this
meeting
is,
as
I
said,
to
just
make
sure
that
those
of
us
who
are
in
you
know
the
sort
of
other
half
of
the
world
to
the
US,
get
a
chance
to
actually
talk
about
this
stuff,
as
opposed
to
just
communicating
asynchronous
secrets.
B
Like
say:
I'll,
be
running,
I
run
this
meeting
every
two
weeks
at
this
time,
so
yeah,
please,
come
and
and
tell
me
what
you
tell
us
what
you
think
I'm
especially
interested
to
hear
from
from
us
sort
of
someone
who's
doing
some
secretly
stuff,
but
is
also
shadowing.
So
you
know
you're
still,
learning
you've
got
you
know.
A
fresh
point
of
view
sounds
pretty
interesting
to
me.
B
Yeah
I
get
up
I
get
up
early
too,
to
go
to
the
US
one
on
early
Wednesday
morning.
Our
time,
but
I
completely
appreciate
that
not
everyone
in
2002
need
to
do
that,
yeah,
so
I
guess
the
main.
The
main
things
that
we're
talking
about
so
at
the
moment
are
just
keeping
everyone
up
to
date
on
what's
going
on,
the
main
things
are
the
surveillance
which
I
posted
in
sand
every
room
as
well,
good,
oh
yeah
and
so
yeah.
If
you
could
encourage
everyone,
you
know
to
use
as
given
a
tease
to
fill
it
up.
B
We
would
very
much
appreciate
it.
We
really
want
to
get
a
nice
big
sample
size
to
be
able
to
draw
some
useful
conclusions.
Then
the
point
of
that
survey
is
just
to
find
out.
You
know
how
people
like
me
about
upgrades.
You
know
how
do
you
upgrade?
How
often
you
upgrade
you
know
what
do
you
like
about
it?
What
do
you
not
like
about
it?
My
gut
feeling
says
that
most
people
will
have
things
they
don't
like
about
it
more
than
things
that
they
do,
but
yeah
just.
We
just
want
to
get
some
feedback.
E
B
Things
and
making
guesses,
so
we
were
just
in.
A
A
loser's
in
just
Kish
I
just
want
to
necessarily
again
there's
a
survey
link
in
the
meeting
notes,
which
you
can
fill
to
express
your
opinion
on
the
current
release.
Cadance
yank,
since
you
have
been
part
of
the
release
team
for
the
last
two
releases,
probably
I
was
also
I'm,
the
shadow
on
the
testin
testin
for
a
team.
So
probably
I,
guess
you
have
good
perspective
on
the.
What
what
how
the
release
should
be.
A
So
you
can
probably
link
click,
a
link
on
that
and
fill
the
survey
and
pass
it
on
to
anyone
you
think
should
and
everyone
who
has
filled.
The
survey
has
told
me
that
they
really
like
this
away
and
the
questions
in
the
survey
really
are
trying
to
address
their
scenarios
with
whoever
they
were
vendors
or
users
or
contributors.
So
you
should
take
a
look
at
the
survey
yang.
A
B
Yes,
so
I'm
just
trying
I'm
just
going
through
the
other
minutes
to
talk
about
it
to
think
about
what
else
to
talk
about
I,
think
yeah,
the
the
main
sort
of
stuff
that
we
had
me
talking
about
is
just
his
small
educators
and
things
that
problems
that
people
have
had,
which
is
kind
of
I.
Guess
an
adjunct
to
the
survey
you
just
sort
of
having
more
chance
to
talk
about
this
stuff,
on
the
assumption
that
the
more
people
know
about
it,
the
more
likely
something
is
to
happen
about
it.
B
Yes,
Oh
personally,
we
we
do
have
s
to
keep
update.
We
are
running
113
in
our
clusters.
At
the
moment,
we
got
delayed
for
quite
a
while
because
of
a
112,
a
bug
that
that
hit
us
really
hard
come
and
what
it
was,
but
but
yeah,
but
we
had
a
whole
bunch
of
migration
work
and
tough
to
change
because
of
the
change
in
113
or
the
demon
sets
now
using
the
regular
scheduler.
And
so
because
we
make
extensive
use
tends
to
taints
and
toleration
x'.
B
We
needed
to
update
a
whole
bunch
of
stuff,
so
that
demon
sets
would
now
be
sure
doing
correctly
by
the
regular
scheduler,
because
you
have
to
tolerate
the
tanks.
I
don't
know
you're
doing
sets
where
you
didn't
need
to
before.
So
that
was
a
good
example
of
the
sort
of
thing
where
it's
a
good
change
for
the
health
of
the
project,
but
it
does
create
a
whole
bunch
of
extra
work
for
us
yeah
one.
B
It's
just
one
of
those
things
where,
if
you
can
yeah,
if
you're
going
to
be
keeping
up,
we
find
that
it
takes
about
a
month
also
of
verification
and
soaking
before
we're
comfortable,
rolling
a
new
version
to
keep
the
prod.
Given
that
we
released
three
monthly
at
the
moment,
that's
like
you
know.
Basically,
a
1/3
of
our
year
spent
painting
the
bridge
of
just
keeping
up
to
date.
A
That
means
that
you
don't
have
to
have
a
stable
version
of
api's
to
release.
You
don't
have
to
wait
for
having
a
stable
version
of
LTS
release,
which
is
an
interesting
proposal.
There
were
lots
of
lots
of
two
and
four.
There
was
out
of
give-and-take
on
the
ideas
and
he
has
taken
up
an
action
item
to
share
that
proposal
after
he
finishes
his
responsibilities
for
the
word.
A
That's
called
that
where
we
can
compare
notes
with
what
Tim
Tim
Sinclair's
proposal
has
and
I
sometimes
feel
that
this
proposals
aren't
competing,
but
there
you
can
make
a
hybrid
version
that
takes
the
best
of
put
the
solutions,
both
the
proposals
or
all
the
proposals
that
get
proposed
and
kind
of
merge
them
into
one
that
solves
the
problems
that
I
survey
brings
out
yang.
Where
do
you
work.
A
E
Percentage
but
many
users
may
have
some
small
clusters,
but
they
are
there
eager
to
you
know
to
know
how
to
properly
upwards
I
mean
if
skip
you
are
running
process
aren't
if
a
year
the
process
will
be
fairly
easy.
But
if
you
are
running,
say,
cops
or
cubing,
the
city
or
some
problem
to
doing
the
process.
C
Right,
I,
that's
not
something:
we've
ever
had
any
trouble
with
we've
been
operating
on
Amazon
for
with
cops
for
almost
two
years
now,
starting
with
whatever
they
currently
supported,
released
for
cops
was
back
then
I.
Remember
what
that
actually
was
and
we've
upgraded
all
the
way
through
we're
now
on
111.
B
B
D
C
B
A
So
we're
your
upgrades,
we're
always
clean.
Did
you
do
in
place
upgrades
John?
Yes,
and
did
you
rely
entirely
on
the
cops
tool
or
did
you
have
wrapper
script
spilled
around
cops
to
verify
test
services
running
on
them?
How
did
you
validate
that?
What
was
running
on
them
will
keep
continuing
to
run.
If
you
have
great
cluster,
how
did
you
validate
that.
A
C
Well,
it's
not,
it
is
on
a
fixed
schedule,
but
it
the
lifetime
is
annoyed
very
somewhat.
We
build
a
new
ami
everyday
and
we
update
all
of
our
lunch
countries
automatically
and
we
automatically
kill
the
oldest
node
every
24
hours,
unacceptable,
asteroids
and
call
them
in
production,
as
well
as
a
way
to
remove
the
OS
patching
burden
for
our
team.
Yeah.
B
You
can
kill
those
nodes
manually
or
wait
for
them
to
drain
away
themselves
with
props,
not
too
bad.
It
works
really
well,
I!
Think
for
us!
Another
thing
that
we
do
for
upgrades
is
what
we
have
the
control
plane
easy
in:
an
auto
scaling
group
and
so
that's
set
to
when
you
update
the
launch
config,
the
AWS,
auto
scaling
group
will
rotate
the
nodes
one
by
one
for
you,
and
so
all
we
got
to
do.
There
is
deploy
the
updated
blood,
config
and
ASG
will.
B
A
Even
if
you're
saying
having
a
stable
release
that
comes
in
after
a
year,
then
you
are
upgrade
every
year
instead
of
upgrading
every
three
months
or
every
month.
How
would
that
look
like
and
how
would
you
be
safely
doing
those
upgrades?
So
it's
a
very
good
example
that
you
have
stated
and
if
you
want
to
take
up,
if
you
want
to
do
on
the
purpose,
we'll
look
at
that
proposal.
That
would
be
awesome
too.
Yes,.
B
Yes,
I
think
you
might,
in
my
mind,
if
you're
running
on
a
clapper
water,
there's
you
may
need,
is
to
be
going
from
keeping
your
notes
available
and
no
never
implies.
Upgrading
anything
that
you
know.
Is
you
Cana
in
this
energy
kind
of
have
the
magic
ice
to
that
you
should
impress
upgrade,
because
in
fly
stock,
relates
to
your
pickle
needing
to
touch
you
in
instances
where,
as
keeping
your
infrastructure,
immutable
means
the
you
know.
If
anything
goes
wrong,
you
just
throw
things
away
and
change
your
confidence
away.
C
The
moment
we
just
do,
we
haven't
really
been
brave
enough
to
build
that
into
any
sort
of
CI.
I
just
tend
to
run
it
myself
with
cops
yeah
on
the
in
place
versus
not
in
placing
it
I
consider
what
we're
doing
kind
of
in
place,
because
we've
still
got
the
same
cluster
data
store
yeah
we're
not
we're
not
doing
the
github
thing,
which
I
would
really
like
to
do
and
build
new
clusters
all
the
time
automatically
performance
testing
like
to
get
there
someday
yeah
yeah.
A
In
Indy,
in
the
you
know
some
of
the
discussions
where
we
call
what
we
call
in-place
upgrades,
it's
very
omastar
and
you're
at
CDs
are
same
nodes.
You
haven't
replaced
them
with
something
else
when
you
do
an
upgrade,
so
you
are
grading
at
CD
two
and
at
CD
2,
HC
d3
and
the
same
machine
by
ticking
down
at
City.
Two
or
you
know
the
machines
are
not
changing.
So
if
you're
changing
the
node
stand,
it
is
not
it's
a
role.
It's
not
a
in-place
upgrade
at
one
point.
A
There
was
core
OS
at
a
proportional
for
self-hosted
kubernetes,
where
you
could
run
multiple
versions
of
kubernetes
at
a
time,
and
that
way
you
can
phase
out
your
services
that
are
running
and
expecting
older
versions
of
services,
apps
or
deployments,
and
you
can
migrate
them
to
newer
versions
of
objects
and
new
versions
of
kubernetes.
So
that's
what
I
think.
So
what
what
do
you
do?
Do
you
use
the
same
nodes
for
upgrades
when
you
do
upgrade
for
at
CDN
master.
C
B
Yeah,
okay,
so
I'm
just
going
to
taking
some
notes
as
well.
Sorry,
but
us
dull.
We
do
we
actually
started
building
a
platform
before
costs,
because
I
think
we
have
our
own
theme.
That
essentially
does
that
same
process
for
like
I,
said
for
the
control
plane
you,
the
ASD
thing
works
great,
but
for
ED
CDs
it
is
a
people
manual.
However,
after
like
years
of
messing
about
with
the
best
ways
to
upgrade
LCD,
the
way
we've
ended
up
finding
is
in
that
you
replace
the
HDD
node
without
actually
replacing
it.
B
So
when
the
new
LCD
node
comes
up,
it's
running
a
new
version
of
it
CD
or
new
iOS
or
whatever
you.
What
have
you
changed
about
it?
But
because
the
IP
address
the
host
name
and
the
data
store
directory
are
the
same
as
far
as
it
CD
the
program
is
concerned,
it's
the
same
node,
and
so
there
was
no
upgrade
done.
B
So
you
know
that
you
know
have
to
worry
about
the
automation
around
safely
ticking
down
the
size
of
the
cluster
as
far
as
a
CD
is
concerned
and
removing
members
and
adding
members
back
in
and
all
that
sort
of
stuff
to
maintain
your
raft
consensus.
As
far
as
raft
is
the
I
know
you
went
away
for
a
little
while
and
then
it
came
back
and
the
fact
that
other
parts
of
the
node
are
different
is
you're,
a
tough
kid
and
I
think
I.
B
A
You
have
a
lot
of
data
points
for
creating
a
proposal
for
how
upgrade
should
happen
say
you
have
stable
or
a
short-term
release.
How
operate
should
be
done
because
I
don't
think
there
is
a
good
pattern.
So
there
are
two
reasons
why
we
need
it
when
we
do
upgrade
testing
in
kubernetes
for
the
short
term
releases,
the
dev
releases
or
the
stable
releases,
we
need
to
define
saying:
okay,
when
you
say,
and
we
say
upgrades
are
supported.
What
do
we
mean
by
operates
or
supported?
A
D
B
I
actually
had
that
talk
accepted
to
keep
on
trying
to
last
year,
but
I
also
had
a
talk
accepted
to
keep
on
us,
and
you
only
got
to
pick
one
travel
budget.
Sadly
so
yeah
I'm
hoping
to
be
able
to
miss
out
on
again
sir
q
Kanchana
this
year.
No
actually
I
miss
the
Q
controller
because
of
financial
years
I
lottery.
Coupon
us
yeah,
traveling
from
everything
from
Australia
to
anywhere
is
expensive
and.
D
A
Five
years
oh
yeah
and
I
think
yang
has
shared
an
upgrade
link,
also
on
one
of
the
talks
at
cook
on
China.
So
yes,
for
instance,
we're
on
at
CD,
John
yang
and
all
of
us
had
some
Cup
points
tomorrow.
Actually,
dev
is
meeting,
it's
not
I,
don't
know
if
it's
a
convenient
time
for
a
pack
zone,
but
CD
dev
is
meeting
tomorrow.
That
will
be
my
second
participation
in
the
meeting.
So
you
guys
are
welcome
to
join
and
like
is
an
Etsy
day
of
coop.
D
A
Now
part
of
C
and
C
F,
so
they
are
following
the
same
model
as
this
meetings,
and
not
many
of
us
would
be
able
to
contribute
to
the
source
of
core
of
fret
CD,
but
we
should
probably
be
able
to
understand
it
is
understand
the
release
Cadence's
and
help
in
the
release.
Engineering
of
it
CD
right
now,
I
think
they
missed.
A
They
have
recordings
true
so
looking
through
their
recordings
junus
when
they
had
decided
to
release
3.4
and
they
come
up
with
a
downgrades
proposal
of
how
it
City
would
downgrade
I.
Think
it's
very
important
for
us
to
discuss
it
in
this
one,
because
we
have
to
also
see
how
the
ecosystem
is
moving
and
how
they
Co
system
is
fixing
and
what
are
the
challenges
if
we
extend
or
reduce
the
cadences
of
a
kubernetes
release,
so
it
City
is
thinking
about
how
downgrades
would
work
when
there
are
up
great
filters.
Kubernetes.
Think
about
that.
A
The
same
way,
the
state
of
kubernetes
is
sitting
inside
the
HCD
cluster,
and
so
it's
important
that
when
you
downgrade
kubernetes
and
if
kubernetes
has
one
point
release
like
it
does
not
say
a
range
of
its
iterations,
it
says
a
particular
version
of
it.
Cd
is
supported,
is
is
tested,
so
we
should
probably
download
kubernetes.
You
have
to
also
Don
great
DHCD,
and
it's
important
at
at
city
is
downgrades
is
supported,
so
3.4
is
coming
out
with
some
kind
of
a
support
and
bug
fixes
for
its
city.
A
A
B
A
Who's
anybody
who
is
doing
at
CD
upgrades
downgrades
themselves
should
look
at
the
proposal
because
they
are
thinking
of
something
as
HCD
downgrades
has
critical
bugs
today
in
that
CD
itself,
when
you
downgrade
at
CD
from
older
versions
to
new
your
minor
and
patch
versions,
you
are
having
some
issues,
so
they
are
trying
to
add
supports
for
both
of
them
and
that's
a
che
at
City
downgrades.
That
means
the
cluster
is
up,
but
the
nodes
are
going
down
similar
to
patterns
Edward.
What
we
have
all
discussed
today,
yeah.
B
Yeah,
that
is
good,
we're
pretty
lucky
now
that
the
tool
that
our
tooling
is
in
a
spot,
where
deploying
an
NC
day
upgrade
is
tedious
rather
than
terrifying,
which
I
would
argue,
is
good,
I.
Think
it's
a
bunch
of
work
like
it's
like
two
days
of
poking
deployment,
things
and
manually
doing
these
actions
in
these
actions,
but
just
like
John
we've
been
a
bit
reluctant
to
made
that
because
it
is
so
critical
to
running
the
cluster.
That
would
rather
have
a
human
in
the
loop
the
whole
time.
B
A
And
I
have
had
bad
times
of
that
city
where
I've
lost
clusters,
so
I
lost
the
HCD
and
I
to
restore
it
from
backups
through
magic
I
mean
acts
of
odd
cities.
I'll
share
two
links,
one.
The
first
link
is
the
it's
a
two
day
meeting
the
the
their
notes.
So
even
if
you're,
not
in
the
meetings,
you
can
see
the
notes.
The
second
one
is
the
downgrade
proposal
that
they
are
talking
about.
A
B
A
B
B
B
Cool
well,
look
I,
really
appreciate
everyone
coming.
This
is
the
first
one
in
a
while
that
I've
actually
been
able
to
have
meeting
and
haven't
just
set
aside
here
for
10
minutes
before
looking
sad
and
closing
the
meeting
down,
and
so
thank
you,
everybody
that's
much
nicer,
so
yeah,
I
guess
I
will
call
the
meeting
here.
The
I'll
upload
the
recording
soon
and
make
it
a
vote
and
put
them
in
the
video
in
the
meeting.
Notes.
B
B
Like
I
said,
we
really
like
both
said
we're
really
trying
really
hard
to
make
sure
that
we
get
as
a
MIT
as
wider
group
of
people
as
possible
to
come
to
contribute
here
so
that
we
don't
end
up
making
something
that
then
all
the
people
who
actually
use
good
real,
you
know
get
crossing
us
about
yeah
and
also,
if
you
want,
if
you
would
like
to
you,
can
contribute.
Please
look
over
the
box
comments
very
welcome
and
pass
pass.
They
passed
the
survey
on
to
everyone.