►
From YouTube: kubernetes kops office hours 20190510
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
Hello:
everyone:
it
is
May
10th
2019.
This
is
cop's
office
hours.
I,
am
your
moderator
facilitator,
just
in
Santa,
Barbara
I
work
at
Google
reminder
that
this
meeting
is
being
recorded
when
we
put
on
the
Internet.
Please
be
mindful
of
our
code
of
conduct
and
be
a
good
person.
I'm
gonna
Pais,
the
notes
there
are.
We
have
an
agenda,
I'm
gonna
paste
that
agenda
into
chat.
A
A
A
B
A
That
I
forgot
about
that
all
right,
yeah
we
should
I
will
have
a
look
at
that
that
we
feed
into
the
conversation
we're
gonna
have
later
around
the
proposal
to
do
some
releases
but
yeah,
okay,
cuz.
We
should
fix
that
I,
don't
know
if
we
really
well
we'll
see
how
bad
it
is
in
terms
of
how
empower
impactful
the
changes
in
terms
of
whether
we
cherry-pick
back
to
112
zero,
which
I
think
we're
about
to
release
so
I
see
you
are
here.
Would
you
like
to
talk
about
your
issue.
C
Sure
can
you
hear
me:
we
can't
okay,
we
are
trying
to
upgrade
from
1.11
to
1.14
and
apparently
it
seems
like
doing
call
cops.
I
did
and
changing
few
parameters
and
doing
doesn't
seem
to
work
for
us
for
whatever
reason,
I
don't
know.
If
I
have
to
remove
some
of
the
flags
in
the
trust
respected,
it
was
generated
when
we
installed
one
anyway.
The
question
is:
is
it
possible
to
run
cops
upgrade
as
part
of
Jenkins
pipeline
that
that's
that's
it
like
the
way
we
are
doing
it
right
now.
C
Is
we
get
the
cluster
machines
cops,
get
and
do
a
cop
side
it
and
then
change
a
few
things
and
then
two
cops
update
on
from
the
local
machine,
which
is
not
really
a
good
practice,
because
we
lose
visibility
on
who
is
great?
What
and
what's
the
outcome
so
yeah
I
was
curious
if
there
is
any
better
way
to
do
that.
Yes,.
A
A
D
C
D
F
In
surprise,
you
do
cox-2
box
template,
we
do
the
same
thing.
We
have
like
a
cellist,
a
first,
a
template
demo,
which
is
in
go
template
and
we
have
value
files
for
each
requesters.
I
mean
we'd,
run
it
as
part
of
a
drone
pipeline,
we're
just
as
cops
template,
cops,
replace
and
then
cops
made
cloud
formation.
Let
me
pass
that
confirmation
to
our.
C
C
A
Validate
is
pretty
like
high
level.
It's
just
checking
that
the
instances
are
basically
there
there.
There
is
a
there
is
a
kubernetes
e
to
e
test
suite
and
a
subset
of
that
is
a
conformance
test
suite,
which
is
believed
to
be
safe
to
run
on
clusters,
and
there
is
a
tool
which
calls
the
criminals
EEE
test
suite
with
the
corrects
flags
which
is
surprisingly
difficult
and
that's
called
Sona
boy,
it's
from
happy.
Oh
now,
VMware
I,
guess
I,
don't
know
if
anyone
actually
uses
son
a
boy
regularly,
but
that
yes,
there
we
go
pretty.
G
E
E
H
Can
also
does
that
we
leverage
of
the
kubernetes
that
there's
the
sweetie
to
write,
custom
and
smoke
tester
and
it
works
pretty
well,
there
was
a
learning
curve
about
one
once
you
know
that
so,
for
example,
we
have
a
cube
2
a.m.
test
ingress
like
just
more
yeah
proper
smoke
test
of
it.
We
run
after
we
upgrade
a
component,
and
that
was
it's
quite
good.
I
C
A
F
Sure,
and
so
they
linked
the
whole
request
there
and
I
think
Justin.
You
commented
on
it
and
in
in
general,
we
have
the
cni
network
mode
for
like
bring
your
own
demon,
set,
we're
actually
using
calico.
So
we
know,
calico
is,
is
supported
by
cops,
but
we
want
to
have
bgp
reflectors
wearing
them
on
a
separate
CD
that
we
don't
learn
and
don't
use
the
cop
set
CD.
F
So
I
made
it
a
fairly
quick,
full
request
to
enable
a
flag
to
following
a
similar
flag
that
we
have
at
the
moment.
But
your
point
is
pretty
valid
by
having
a
full
kind
of
security
group
style
would
be
awesome,
I'm
just
going
to
be
selfish
and
be
like
I
think
the
IP
IP
might
be
the
most
common
use
case,
but
yeah
I
understand
that
making
it
generic
can
awesome
would
be
the
best
way
of
going
about
it.
So
just
getting
people's
ideas.
A
Yes,
I
so
I
think
yet
to
summarize
like
shy,
quick
Alex
or
tomorrow,
which
is
your
preferred
that
makes
this
one
okay
Alex
proposed
to
PR.
We,
we
added
a
field
to
the
CI
networking,
spec,
just
called
I,
think
aloud.
Protocol,
IP,
IP
and
I
sort
of
suggested.
That's
actually
really
cool,
and
we
definitely
have
this
problem
everywhere.
Where
you
know,
how
do
we
separate
out?
A
We
eventually
want
to
get
all
the
CNI
providers
and
everything
separated
out,
but
there
is
this
cross
dependency,
where
different
CI
providers
need
different
ports
or
IP
IP
open,
and
can
we
somehow
specify
that
and
I
I
proposed
that,
like
we
borrow
the
network
policy,
syntax,
there's
a
lot
of
Nitra
policy,
syntax,
but
I
I,
don't
know
if
we
can
just
do
a
subset
of
it.
So
it's
not,
as
you
say,
like
acts
like
a
massive
effort
like
I.
F
My
question
would
be
how
to
specify
the
different
nodes
roles,
so
I
get
a
master
or
worker,
because
in
this
case
it's
kind
of
just
a
cross-product
in
the
firewall
code,
which
kind
of
says
here
all
of
the
Masters
aSG's
and
all
of
the
work
at
SGS
can
trust
each
other,
and
that
becomes
quite
verbose.
If
you
have
to
put
in
yellow.
A
Yes,
I
think
so
my
understanding
is
this
would
be
in
the
network
policy
spec,
but
I
just
didn't
copy
that
bit.
So
if
we
wanted
to
do
I
think
we
should
default
to
it
being
you
know
the
cross-product
but
I
think
the
the
first
link
that
I
gave
includes,
like
a
from
in
the
network
policy
ingress
rule
which
I
then
renamed
noting
rest
rule,
but
network
policy
ingress
rule
has
a
from
which
is
like
says
that
some
things
can
like
that
determines
like
they
use
pod
selectors.
So
we
would.
A
A
A
A
F
A
A
All
right,
so
next
up
I
am
going
to
propose
that
we
are
more
or
less
ready
to
release
cups,
112,
0,
&
113,
a
cup
1
table
0.
Let's
start
there
and
Ryan.
Thank
you
to
Ryan
who
did
a
lot
of
work.
Helping
me
figure
out
all
the
cherry
picks
that
we
needed
it
looks
like
there's
a
comment
on
Peter
I:
don't
we
move
the
spreadsheet
around
so
I'm,
not
sure
where
that
comment
is.
Do
you
remember
where
that
comment
was
you
mentioned.
I
A
I
A
Legend
absurd.
Thank
you.
Thank
you
all
right.
Yes,
so
we
have
this
spreadsheet,
which
is
linked
in
the
agenda
for
people
that
want
to
click
on
it
and
we
have
gone
all
through
112
and
I.
Think
everything
has
been
cherry
picked.
These
things
that
have
there
were
about
five
or
six
that
have
not,
which
are
cleanups
and
refactorings,
which
we
probably
don't
need
unless
a
future
PR
needs
them.
So
I
think
the
right
thing
to
do.
A
Is
we
only
cherry-pick
those
if
a
future
cherry
pick
needs
some,
they
shouldn't
be
I,
don't
anticipate
that
we
will
actually
need
them.
So
if
I
think
there
was
one
very
valid
objection
brought
up
earlier,
which
is
the
a
diverse
region,
bug
I,
don't
know
if
anyone
else
has
any
blockers
on
1:12
that
would
block
1:12
zero
going
once
twice
three
times
done,
just
kidding
the
yes.
If,
if
you
do
then
yeah,
please
feel
free
to
speak
up.
It's
a
it's
overdue.
G
F
C
G
I
think
one
thing
maybe
we
I
would
propose
is
if
we
can
fix
that,
you
know
it's
a
PR,
so
you
know,
maybe
maybe
we
can
merge
that
or
if
there's
adjustments
needed,
you
know
try
to
get
those
done
ASAP,
but
if
not
like
hold
up
the
release,
but
only
so
long
you've
got.
It
then
set
the
release
and
if
we
really
need
to
create
a
you
know
a
112
one
that
make
sense.
That's
just
my
two
cents,
yeah.
B
A
Okay,
that
works,
and
then
I
was
also
proposing
that
we
we
bump
our
new.
We
have
113
and
114
in
flight
as
weak
as
it
were,
so
we
can
then
do
113
0,
beta
1
and
114
0,
alpha
2,
basically
getting
those
rolling
forward,
so
that
we
are.
This
is
part
of
our
big,
like
catch-up
effort,
which
is
you
know,
the
the
upside
of
all.
Those
cherry
picks
with
the
crazy
spreadsheet
is
that
we
can
actually
hopefully
catch
up.
Maybe
not
my
cube
con,
but
pretty
soon
afterwards.
B
A
A
I
will
do
that.
Yes,
I
would
do
that.
It's
right,
I'm
not
traveling
today,
so
it's
it's
possible
or
more
practical,
and
then
we
have
two
items:
I
moved
to
the
bottom
of
the
agenda
about
canceling
next
meeting
or
not
canceling.
The
next
meeting,
which
I
thought
was
amusing
to
put
those
right
off
against
each
other.
So
I
was
suggesting.
We
should
cancel
the
next
meeting,
which
is
May
24th
in
honor
of
you
kind
of
you,
I
think
technique.
You
con
ends
on
Thursday,
but
I.
A
A
G
H
G
C
A
A
F
A
A
If
there's
nothing
else,
I
thank
everyone
for
their
time
and
for
all
their
wonderful
PRS
and
issues,
and
thank
you
and
hope
to
see
as
many
of
you
as
possible
in
Barcelona
and
it
sounds
like
we
will
not
do
cops
office
hours
on
the
Friday
in
two
weeks,
the
24th,
but
we
will
instead
do
it
next
on
the
7th
of
June
and
we
will
hopefully
have
a
cop's
office
hours,
on-site
lunch
or
off-site
lunch
in
Cuba
Connie
here.