►
From YouTube: Kubernetes Kops Office Hours 20180914
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
A
B
A
Yeah,
thank
you
guys
right.
It
was
just
because
I
was
sharing
the
screen
to
show
it.
I
was
starting
to
find
a
beautiful,
the
I
think
if
we
were
able
to
turn
on
a
CD
manager
by
default,
which
I
think
we're
able
to
do.
There
are
definitely
some
issues
but
I
think
we're
able
to
turn
it
on
by
default
in
release
and
I.
Think
that
should
be
these
my
goal
for
111
and
it
might
determine
timelines.
A
But
if
we're
able
to
turn
on
and
see
you
manage
my
fault,
then
I
think
we
should
make
at
CD
3
the
default
for
new
clusters
and
yeah.
So
we
should
make
it
two
clusters,
the
only
people
of
how
many
people
the
people
left
on
at
CD
2
will
be
everyone
that
has
an
existing
cluster
and
the
intention
is
that
it
is
easy
and
verified
that
you
can
update
at
CD
3
schedule
and
what
you're
not
forced
to
do
so
until
I.
Think
113
is
gonna,
be
the
one
that
deprecates
it.
A
Yes,
it
will
be
forced
on
and
it
will
be
a
forced
upgrade
at
that
point.
So
the
the
current
blocker
I
think
there's
some
minor,
relatively
minor
issues.
I,
don't
know
any
correctness
issues.
Yet
the
the
issues
are
an
H.
A
rolling
upgrade
has
a
only
a
one
in
three
chance
of
working
because
you
have
to
because
it
depending
on
which
one
is
killed
depending
on
sort
of
one
sorry,
one
I'd
CD
1h
a
upgrade
path.
A
So
when
you
turn
on
the
manager
first
and
then
go
front,
CD
de
83
is
unlikely
to
work
without
a
force
on
your
rolling
upgrade.
So
we
need
to
either
just
document
that
or
come
up
with
a
workaround.
That's
just
to
do
a
sort
of
the
leader
and
refresh
like
refresh
cycles
on
status
or
I
need
to
think
about
that
the
much
bigger
one!
That's
not
that's,
not
a
huge
one.
A
The
much
bigger
one
is
calico
is
currently
talking
directly
to
at
CD
and
there's
a
third
ones,
all
cook
we
talking
directly
to
CD,
and
so
we
need
to
integrate
calico
into
the
upgrade
in
some
way
and
I
think
probably
the
best
way
to
do
that
is
actually
to
to
switch
calico
to
use
their
new
or
API
driven
one.
So
they
don't
look
at
CD,
which
I
think
will
make.
That
is
a
better
approach.
A
It
means
we
don't
have
to
open
at
City
of
the
world
of
freezing
sorry
open,
SDG,
your
nodes,
if
you're
using
calico
so
I,
think
that's
great
and
then
othered
the
other
thing
that
is
TLS
support,
which
we
need
to
get
working
as
well.
So
I
think
that
is
my
goal,
whether
that's
well
feasible
for
the
111
timeline.
I,
don't
know,
but
I
would
like
to
try
to
happen.
Make
it
happen,
we'll
see
what
else
is
eleven
so.
A
B
A
A
A
So
that's
sort
of
the
way
it
was
sort
of
the
the
sequence
again
got
it
and
that
the
sorry,
the
the
sort
of
the
the
forcing
function
is,
if
113
he's
going
to
turn
off
at
CD
3,
it's
gonna
turn
off
at
CD
2
and
we
sort
of
need
to
like
be
making
progress
towards
getting
that
forced
on
people.
So
we
have
enough
mileage
on
yeah.
B
A
A
B
See
me
yeah.
This
is
one
of
my
other
issues
as
well.
I
guess
we
can
probably
talk
about
this.
This
is
one
of
my
other
agenda
points
as
well.
Is
this
road
map
for
at
CD
manager,
which
is
out
of
date
and
I,
was
wondering
if
we
should
converge
it
with
with
the
main
road
map?
Although
you
know
I
I,
guess
that
bei,
it's
probably
gonna,
live
on
as
a
separate,
separate
concern,
and
so
maybe
it
needs
to
be
separate.
Yep
there
remembers
about
the
other
day.
Okay,
yes,
I,.
A
A
They're
no
longer
Road
map,
yes
and
they're.
Not
that's
a
look,
oh
yeah,
because
that's
easy
through
that.
If
all
is
not
yes,
correct
yep,
so
we
didn't,
we
didn't
make
it
to
be
through
the
default.
Yet
alright,
I,
yeah,
okay,
I'm,
not
sure
whether
I
think
sed
backup
is
easily
enabled
button,
still
not
the
default.
But
yes,
cool.
B
B
We
definitely
need
a
bunch
of
work
on
our
documentation.
I
think
that
it
needs
to
be
like
odd,
because
there's
a
lot
of
stuff,
that's
old
or
not
relevant
or
things
are
like
inconsistent.
We
different
places
or
I
think
we
have
a
lot
of
documentation
now,
which
is
very
different
than
when
I
started
working
with
cops.
But
it's
like
too
much
almost
and
it's
not
it's
very
difficult
to
use.
I
think
so.
B
A
B
A
B
A
A
A
A
B
I
know
I
do
fit
your.
Maybe
we
can
have
just
sort
of
like
a
section
for
you
know,
general
ideas
that
we're
working
to
incorporate
but
are
not
tied
to
any
specific
release,
because
there's
a
few
things
like
I
want
to
rewrite
the
networking
spec
or
make
a
networking
inspect
for
a
networking
object.
I
think
that
would
help
us
solve
some
of
the
problems
that
I
find
offensive
in
your
network
inside
and
then
also
yeah
things
like
cops
server,
machine,
API
stuff
like
that.
So
I'm
happy
to
pull
that
into
like
a
separate
section.
A
B
Cool
that
sounds
great
I'm
happy
to
I
can
finish
this
off
and
submit
a
new
version
now
this
afternoon.
Thank.
A
E
A
A
I
think
the
the
challenge
is
gonna,
be
the
work
that
I
would
like
to
happen
on
things
like
get
city
manager
and
bundles,
and
where
that,
where
we
slap
that
in
yeah,
so
maybe
we
put
Exedy
manager
into
one
well,
we
have
we
have
a
team
under
support
today.
Maybe
we
make
it
the
default
in
112,
so
we
don't
yeah.
B
So
I
think
that
we
should
I
think
that
we
need
to
move
quickly
and
one
get
something
out.
I
think
we
should
try
to
catch
up
in
my
opinion
and
and
I'm
I'd
be
interested
to
see
what
folks
have
to
say,
but
I
think
it's
useful
to
catch
up.
Even
if
we
don't
provide
every
single
feature
like
right
off
the
bat.
It's
it's
hopefully
catch
up
with
a
reason
really
cycle.
So
people
aren't
trying
to
like
hack,
cops
to
make
your
stuff
work.
D
A
A
Like
they
did
catching
up,
but
I
think
that
makes
sense.
So
we
will
try
not
to
put
too
much
into
111
yeah,
okay,
the
the
yes
I
think
they.
The
interesting
thing
has
been
the
MV
image.
Version
of
EDD
manager
is
sort
of
awkward
without
bumbles.
So
that's
been
one
of
the
things
that
I've
been
worried
about
that
we
can
put
in
a
big
release
note
in
for
people
that
pre-adult
@cb
manager
before
bundles.
If
we
put
bundles
into
112,
cops
112.
B
Because
it
is
it
totally
unfeasible
to
I,
mean
I,
mean
I.
Think
I
think
it's
definitely
preferable
right
now
to
sort
of
try
to
stick
to
some
sort
of
regular
release
cycle,
and
you
know,
try
to
act
a
little
bit
more
like
the
regular
like
like
kubernetes
upstream
repo,
where,
if
you
get
your
feature
and
you
get
it
in,
if
not
like,
we
don't
hold
up
release,
but
for
that
feature
obviously,
but
is
there
a
way
that
we
can?
You
know
kind
of,
do
like
a
point
release
like
a
patch
release.
B
A
B
A
A
A
We'll
see
what
this
data
set
bundles
is
because
it
would
be
nice
to
have
to
have
a
way
to
sort
of
be
able
to
get
the
functionality
of
bundles
without
for
STD
manager,
with
its
to
avoid
the
sort
of
luck
and
the
issue
is
right
now.
The
way
you
turn
on
EDD
manager
is,
you
think,
actually
that's
not
true,
I
think
so
right
now,
the
only
way
to
get
a
new
HCD
manager
image
is
to
update
cops
that
will
be
fixed
by
bundles.
Today
we
could
fix
it
using
channels,
but
we
don't
really
want
to.
A
We
don't
do
that
it's
sort
of
without
bundles,
in
which
the
version
would
be
unpinned,
and
it
would
you
sort
of
automatically
get
updates
whenever
it
changed
or
whenever
you
next
didn't
on
reddit
cops
operation.
So
that's
a
little
bit
of
a
weird
functionality.
We
can
talk
about
that
more
in
the
bundle
section
it
looks
like
we
have
one
more
coming:
she's
Chris.
You
want
talk
about
that.
Cd
manager
and
operators
status.
A
A
A
A
C
A
A
So
it's
only
calico
and
flannel
does
not
sorry
canal
does
not
doesn't
have
a
special
exception
Oh,
although
I
guess
it
could
be
talking
to
it
directly
on
the
master
I.
Don't
think
it
is.
But
anyway
that's
a
good
question.
I,
don't
think
I,
don't
think
that's
the
case,
but
it
it
may
be.
I
think
you're
right,
I've.
A
A
We
have
to
I
just
wanna
show.
Well,
because
the
manifest
is
bed
in
two
cups.
We
would
have
to
do
a
new
release
of
cops.
The
we
have
something
called
channels
which
is
a
sort
of
file
where
we
get
around
this
4:00
a.m.
eyes
and
kubernetes
versions
where
we
were
able
to
recommend
those
without
doing
any
race
of
cops.
A
The
idea
of
a
bundle
is
the
idea
of
taking
a
concept
like
channels
and
having
a
manifest
file
that
includes
all
the
add-ons
in
it
and
then
we're
able
to
pull
out
the
add-ons
that
we
need,
and
the
idea
is
that
then,
that
manifest
can
be
released
separately
from
cops,
and
so
you
can
get
a
fixed
version
of
calico
without
having
to
get
a
new
version
of
cops.
So
when
calico
does
a
fix,
they
can
we
can
do
a.
A
A
The
some
people
from
gke
have
released
a
specification
which
is
based
on
what
they
do
inside
gke
sort
of
some
of
their
learnings.
It's
a
Google
cloud
platform
case
cluster
bundle.
So
if
we
use
this
format,
the
idea
is
that
then
we
won't
have
to
do
it
all
on
our
own,
but
rather
will
all
be
able
to
share
a
bundle,
a
set
of
a
set
of
manifests
for
our
add-ons.
A
So
when,
if
your
imagine
that
you're
a
calico
developer
and
you
update
a
new
version
you're
not
to
go
update
the
cops
repo
and
the
cuvette
TM
repo
and
kubernetes
anywhere
repo
and
the
cluster
api
repos
that
are
springing
up
you,
but
rather
you
just
update
something
which
holds
the
bundle
and
then
everyone
sort
of
picks
it
up
from
there.
So
including
GK
I
guess.
A
Obviously,
there
will
be
some
gates
before
going
straight
into
GK
II,
but
you
know
that's
sort
of
how
the
ideas
that
then
everyone
in
the
community
community
will
have
a
nice
way
to
manage
their
add-ons
or
the
declaration
of
your
add-ons.
That's
the
bundle,
and
so
the
idea
of
many
of
this
PR
is
that
we
start
using
the
bundle.
A
A
When
you
do
a
like
unrelated
change
to
your
cluster,
like,
for
example,
if
you
were
to
change
from
m3
mediums
to
m4
mediums,
you
don't
want
to
like
suddenly
pick
up
a
new
version
of
calico,
and
so
the
idea
is
that
spec,
the
cluster
spec
adds
a
new
field
which
thank
you
first
thing
for
that.
The
cluster
spec
adds
a
new
field
called
bundle
which
is
sort
of
a
particular
version
of
the
bundle,
and
that
determines
the
versions
of
all
the
add-ons.
A
So
then,
what
we'll
do
is
we'll
have
the
ability
to
sort
of
keops
create
cluster
will
automatically
pick
up.
The
latest
bundle
from
the
channels
file
that
we
still
use
and
cups
upgrade
will
recommend,
and
if
you
say
yes,
we'll
automate
updating
the
bundle
so
updating
the
version
of
the
bundle
that
you're
using.
B
A
A
The
I
guess
a
future
world
would
be
that
calico
publishes
like
their
portion
of
the
calico
bundle
or
their
their
add
on
their
their
calico
specification
as
a
bundle,
and
then
we're
able
to
sort
of
aggregate
it
up
and
then
a
sort
of
future
future
version
is
that
you
know
when
cops
and
cube
anywhere
and
all
the
other
tools
are
all
using
are
all
using
the
bundle.
Then
we'll
put
it
the
bundle
in
somewhere
that's
outside
of
any
one
particular
any
one.
Particular
installation
tool,
foster
management
tool.
B
A
Not
currently
the
case,
yes,
we
could.
We
certainly
could
the
bundles
earlier
than
that
so
I'm
thinking
that
we
start
by
using
just
for
EDD
manager.
Oh
yes,
we're
gonna
have
to
have
either
different
bundles
for
each
kubernetes
version
of
our
range
of
communities,
versions
or
we're
gonna
have
to
have
in
a
particular
bundle
nibble
a
to
specify
the
particular
manifests
apply
to
particular
kubernetes
versions.
So
today
that's
more!
We
what
we
have
and
the
primary
reason
why
we
have
that
well,
they're,
actually,
two
reasons.
The
primary
reason
why
we
have.
A
A
The
the
other
issue
is
that
we
don't
necessarily
want
to
introduce
a
new
new
functionality
to
an
IKE.
We
don't
want
necessarily
run
the
latest
calico
on
a
one
six
cluster,
if
we're
not
testing
that
that
newer
version
of
calico,
so
that
is
a
trickier
problem
and
that
that
might
be
solved
by
pinning
to
an
older
version
of
the
bundle
or
it
might
be
solved
by
having
the
bundle
have
both
versions
and
I.
Don't
you
don't
think
where
you
know
what's
gonna
happen?
A
Yet
the
the
usage
that's
proposed
here
is
just
to
have
a
CD
manager,
which
is
actually
just
a
pod
that
actually
it's
rendered
as
a
static
pod,
have
the
manifest
for
a
CD
manager,
so
very
stable.
Have
that
be
in
the
bundle,
rather
than
being,
it
seems
like
a
very
safe
place
to
start
and
I'm
hoping
we
can
start
some
of
like
coordinates,
for
example,
would
be
a
good
one.
Yeah
course
the
default
yet
so
that'll
be
a
great
one
to
put
into
the
bundle
as
well.
This
is.
D
A
Yeah
I
think
I
think
the
same
concept.
I,
don't
think
that
there's
as
much
I
don't
gives
any
need
to
bundle
a
sari
to
version
the
it
was
mission
database
database
I
was
thinking
looking
at
the
PRS,
as
you
saw
this
morning,
I
presume
it
would
be
great
to
get
that.
I
was
thinking
that
could
be
a
new
project
under
the
sig
AWS.
A
That
can
be
a
fairly
small
like
a
simple
project
under
sagaie
WS,
where
we
basically
build
a
llamó
file
up
all
those
machine
types,
I,
don't
I,
don't
know,
I,
don't
know
any
reason
why
you
see
versioned,
maybe
you
do
but
yeah
so
I
think
the
same
idea
is
a
bundle
where
we,
you
know,
define
a
spec
ferrah
cool,
have
some
tooling
to
create
it
and
publish
it
in
a
well-known
location.
So
in
other
words,
this
would
sort
of
it's
the
same
same
evolutionary
path
right
as
we're
talking
about
here.
We're
just
going
straight.
A
Sure
I
think
so
the
one
thing
that's
so
one
thing
that's
annoying
about
this
proposal
is
today:
if
you
do
a
new
version
of
cops,
you
automatically
get
the
new
version
of
calico
that
we
bundled
in
we
that
we
shipped
with
the
new
version
of
cops.
If
we,
if
we
have
a
bundle
and
we
pay
into
a
bundle,
you
will
not
get
the
new
version
of
calico,
surely
by
upgrading
cops,
you
have
to
update
the
bundle
field
we
can.
B
F
I,
do
I
see
many
people
who
don't
know
how
to
actually
edit
the
cops
back
and
everything
who
will
get
stuck
with
it
just
because
if
they
they
used
to
run
cops
up
through
it
cut
cops,
update,
cops,
upgrade
cluster
and
they're
done
so
I
see
kind
of
the
issue
with
people
with
no
idea
we're
just
playing
around
with
it,
like
Isis,
really
preparing
trainings
these
days,
we're
just
like.
Oh,
we
just
fill
up
a
cluster
with
cops,
which
is
great,
but
then
there's
people
know.
Oh,
it's
like
three
commands
and
everything
works.
A
Say
yes,
and
we
can
that's
a
good
point
and
we
can
so
it.
It
is
already
integrated
in
this
PR
into
clock.
Ops,
cluster
upgrade.
So
just
as
that
was
like,
say
hey,
you
should
use
an
AMR
a
new
AMI
or
we
recommend
a
new
kubernetes
version.
It
will
now
operate
a
new.
It
will
now
recommend
a
new
bundle
version
and
apply
that
change
for
you.
A
If
you
say
yes
and
then
the
the
other
thing
we
can
do,
which
we've
reserved
in
the
past
for
like
pretty
serious
like
a
security
issue,
type
issue
in
communities
is
we've
said:
hey
this
version
of
kubernetes
is
really
not
recommended.
We
we
will.
We
exit
cops
unless
you
pass
a
magic
environment,
verbal
which
says
I
know
what
I'm
doing
so.
We
could
do
the
same
thing
with
the
bundle
where
we
say
like
required.
A
I
think
as
we
call
it
recommended
version
require
a
version,
and
if
you,
if
you're
running
an
older
version
of
the
bundle,
then
we
require
we
would
we
could
do
the
same
thing
where
we
say
this
is
a
serious
enough
flaw
that
we
really
fail.
The
challenge
is
that,
just
because
calico
has
a
flaw
doesn't
mean
that
everyone
else
in
the
world
should.
F
Have
to
upgrade
I
can
actually
think
of
like
one
command
to
just
go
to
the
mobis
latest
recommended
version
like
just
update,
expect
to
the
latest
kvitko
whatever
to
latest
kubernetes,
and
everything
like
when
you
just
see
the
arrow.
You
still
don't
know
what
to
do
and
we,
like
the
documentation,
doesn't
really
easily
tell
what
to
do
as
well.
So
maybe
like
cops
update,
something
like
cops
go
to,
recommend
it
or
something.
I
would
probably
make
it
easier
for
lots
of
people
also
by
not
looking
up.
A
So
that
I
mean
that
is
sort
of
what
cops
cluster
upgrade
is
supposed
to
do,
but
yes,
we
should
certainly
make
that
more
discoverable
and
we
can
sort
of
say
we
have
cop.
So
there
is
cop
set
support
for
this
and
we
can
say
well
if
we,
if
we
detect
an
older
version,
we
can
say
here
are
the
two
recommended
actions
you
can
cups
cluster
upgrade
or
you
can
cop
set
cluster
dots
backed
up
bundle
equals
x
and
we
would
actually
populate
x.
That
could
be
good
I'll
write
that
down
as
oh.
F
B
I
think
I
think,
there's
some
good
points
coming
out
here
relating
to
like
workflows,
and
I
think
that
that's
something
that
I
like
kind
of
what
I'm
CUBAN
Eddie's,
that
iOS
done
with
their
Doc's
over
the
last,
like
version
review,
where
they've
created
like
sort
of
use,
cases
and
stories
that
you
work
through
and
I,
think
that
we
should
think
about
restructuring
our
Docs
in
that
way.
And
then
this
and
kind
of
create
a
list
of
these.
These
types
I.
A
Think
they'll
be
great
I.
Think
if
you
have
a
bit
I
think
I
did
some
GCE
docs
that
are
sort
of
more
along
those
lines
and
I
think
I'd
be
great
I,
think
I
wasn't
I
haven't
yet
been
brave
enough
to
try
to
tackle
it
for
AWS.
Where
that
you
know
the
journey
is
much
more
there's
a
lot
more
complexity
in
the
networking
lot
more
options,
but
yeah.
A
If
you
want
to
have
a
look
at
the
GCE
ducts
I
think
you
know
if
you
want
to
try
to
take
on
for
a
device
that
be
amazing
and
doesn't
need
to
be
that
different
right,
hopefully,
most
of
that
most
of
the
similarities
in
nativist
most
of
the
differences
most
of
the
workflow
is
specific,
at
least
in
the
beginning,
rather
than
AWS
versus
GCE
like
it's.
How
do
you
create
a
cluster?
How
do
you
upgrade
a
cluster?
How
do
you
resize
it?
How
do
you
delete
it?
B
C
A
I,
did
it
that
way?
Okay,
if
I'm
nice
to
have
I
don't
know,
have
something
more
then
I
single
out
dry
run.
Is
it
actually
like
you
can
use
it
to
output
Y
Amal,
which
is
nice,
whereas
cops
buster
upgrade
doesn't
and
just
has
the
weird
yes
or
absence
of
yes
and
then
just
does
it
coffee
cluster
upgrade
there
doesn't
apply
your
change,
it
doesn't
actually
update.
It
doesn't
do
a
couple
update,
so
some
of
this
confusing
rich
boys
after
cups
cluster
upgrade.
You
still
have
to
do
a
cups
cluster
update
the.
A
A
C
Yes,
in
effect,
we
we
keep
the
Yambol
in
a
Jinja
to
template
that
we
drive
in
that
we
run
update,
f
la
blah,
blah
off
of
that
file.
So
I
do
whenever
there's
cops
upgrades
or
whatever
major
structural
changes
happen.
I
need
to
run
up
a
dummy
cluster
of
the
to
get
to
export
the
new
gamal
and
do
a
diff
comparison
and
blah
blah
blah.
So
something
to
streamline
that
would
be
appreciated.
Yeah.
A
A
B
Can
just
go
quickly
on
this
next
next
data
about
the
outreach
e
of
proposal
Paris
solicited
she
put
out
a
call
for
for
mentors
for
the
it's
gonna,
be
like
the
google
Summer
of
Code,
but
it's
in
the
winter
and
it's
gonna
be
sponsored
by
outreach,
e
and
I.
Think
Chris
and
Mike
were
the
ones
that
spotted
that.
So
thank
you,
but
I
was
thinking
that
we
should
get
some
helpers
or
a
helper,
and
that
would
be
a
good.
B
You
know
good
way
to
get
more
visibility
for
cops,
but
also
to
get
some
help
on
some
of
these
things
and
I
I
can't
imagine
that
people
are
going
to
say
no
to
this,
but
we
do
need
to
create
a
proposal.
There
is
some
sort
of
we
have
until
the
28th.
So
we've
got
two
weeks
from
today
to
put
this
together.
I
think
and
I
think
I'd
like
to
get
some
some
people
to
get
together
and
put
their
ideas
down
in
a
doc
and
and
and
we'll
get
it
off.
Is
that
I'm
animal.
B
F
Mean
it
can
really
be
anything
right,
so
the
support
could
be
if
the
project
says
develop
this
and
that
features
cops
like
you
should
really
create
a
project
and
the
project
says
create
a
future.
It
probably
makes
it
make
sense
to
people
give
them
the
requirement.
You
know
go
if
you
say
well,
it's
mostly
about
documentation
which
we
could
probably
need
a
little
more
lots
more
help
as
well.
I
would
also
make
quite
good
sense
of
table.
They
just
need
some
technical
expertise
to
just
go
through
it
and
write
stuff
down.
F
We
just
need
to
think
of
a
project
and
like
just
spec
that
out
create
some
issues
in
I
think
in
the
project
and
label
them
with
good
first
issue
that
people
that
can
actually
use
that
too
I
gotta
drive
to
like
apply,
so
they
try
to
solve
something
in
order
to
apply
for
the
project.
On
the
other
side,
of
course.
Well,
someone
needs
to
pay
these
people.
So
if
any
of
your
companies
has
like
six
point,
five
grand
to
spend
I
think
feel
free
to
do
so
outreach.
F
He
is
some
kind
of
nonprofit,
so
it
helps
you
with
their
tax
as
well.
If
you
want
to
do
it
privately,
I
think
anyway.
So
it's
just
like
getting
some
help
and
maybe
also
some
visibility.
I
know
the
CIA
Isaac
is
going
to
participate.
Paris
is
pushing
it
quite
bad,
but
yeah.
If
no
one
is
actually
like
I
think
we
have
enough
mentors.
We
just
need
to
spec
it
out.
Yeah.
B
I
think
that
I've
started
to
you
and
whenever
I'm
I've
been
doing
some
issue
triage
and
such
and
trying
to
drop,
drop,
good
starter
issues
on
things,
that'll
get
get
us
going,
at
least
in
that
side.
But
please,
when
you're
writing
issues
like
if
it's
something
that
makes
sense
to
take
it
that
way,
so
that
it's
easy
to
find
I
think
as
far
as
documentation
like
I
think.
B
That's
not
like
the
best
I
mean
I
would
love
to
have
somebody
help
us
with
documentation,
but
I
think
it's
gonna
be
a
tough
task
for
some
I
think
it
might
be
a
very
tough
task
for
somebody
that
was
pretty
junior
because
there's
a
lot
of
like
sort
of
like
structural,
like
on
editorial
choices,
that
need
to
be
made
I
think
is
that
that,
like
I,
think
we
need
like
major
changes.
Not
little
changes
in
Docs,
but
I
do
think.
B
B
So
we
we
have
to
I
think
there
are
two
basic
pieces
to
this.
We
wanted
to
we
yeah.
We
need
to
create
some
sort
of
short
list
of
projects
or
tasks
that
we
could
have.
Interns
work
on.
The
other
is
to
have
something
so
that
people
that
are
interested
in
applying
for
the
project
have
a
way
to
get
their
foot
in
the
door.
So
it's
a
good
starter
issues,
places
to
just
get
their
name
in
there.
B
D
B
A
G
Yeah,
so
it's
pretty
much
the
one
you
created
blissful
and
Chris
created,
plus
the
EOB
security
group
changes.
So
it's
massive
thanks,
Chris
and
Eric
for
taking
a
look
at
it.
So
quick
thing
Justin
after
talking
to
Eric,
we
kind
of
want
some
guidance
from
you.
One
of
the
things
that
we're
not
super
happy
about
is
because
we
already
have
additional
security
groups.
G
Then
the
secure
groups
that
we
create
we're
wondering
if
there's
any
way
to
clean
this
up,
so
we
don't
have
yet
another
spec,
that's
out
there
that
define
security
groups
that
people
can
get
confused
with.
So
we're
wondering
if
there's
a
way
to
consolidate
that
of
those
I,
couldn't
think
of
any
clean
way
to
do
it,
I'm
wondering
if
you
can
I.
A
Mean
I
think
I.
Think
additional
security
groups
is
a
is
a
nice
thing
to
have
right
additional
security,
which
is
when
you
want
to
give
more
permissions,
yep
and,
and
so
to
me
they
are
different.
This
is
like
using
additional
security,
which
is
basically
pretty
safe.
Mm-Hmm
security
group
is
gonna,
be
pretty
yeah
in
it.
I
like
it.
Maybe
yeah
security
group
yeah,
the
security
group,
security
group,
override
yeah
and
then
I
think
whether
we
bother
having
the
word
override
in
there.
A
A
B
E
A
I
mean
I,
just
I'm
I'm,
actually
happy
with
security
group,
I
think
but
yeah
we
can
do
primary
security
group
or
overrides.
If
your
group
is
clear
as
to
what
we're
actually
doing
and
that
it's
it
is
actually
very
like
it
also
seems
dangerous,
whereas
digital
it
seems
like
yeah,
you
know
put
more
on
them
more
on
the
pile.
Where
is
it
override
is
clearly
like
I
am
I'm,
bypassing
cops
mm-hmm.
E
A
A
G
B
G
A
B
Sorry,
oh
yeah
I
would
say
sometimes
it's
nice
for
them
to
have
that
warrant,
because
we
can
pick
that
up
very
quickly.
If
we've
seen
you
know
a
bunch
of
specs
that
are
broken
like
that.
But
having
somebody
say
like
just
having
that
hints
like
it,
sometimes
really
helpful
for
people
that
don't
have
a
good
like
holder,
where
they're
going
and
then.
A
Again,
we
could
also
put
it
into
cops
valid
a
if
you're
cops
mounted
a
cluster
fails.
We
could
be
like
here
are
the
things
that
you're
doing
that's
a
little
bit
out
of
the
no
ordinary
right
you're.
That's
yeah,
you're
messing
with
your
security
groups.
You're
like
messing
with
you,
I
am
you're
you've
overriding
overridden
the
image.
You
know
anything
like
that,
that's
even
it
it's
just
like
you
know
here
are
the
things
that
we
think
are
the
most
unusual.
Things
are
doing
that.
That
could
be
helpful
for
yeah.
B
A
D
A
Six
I'll
just
to
prove
this
missus
I,
know
you're
some
reason,
but
I
want
to
prove
this
and
the
rules.
You
know
everything
was
just
updating.
Yes,
we
should
actually
update
those
anyway.
I
didn't
want
to
update
it
just
before
the
release.
But
yes,
it's
time
to.
We
should
update
it
for
111
and
he's
not
young
as
soon
I
reach.
A
woman
was.
D
A
D
A
I,
don't
know
how
many
voters
still
building
with
make
versus
basil
I
suspect.
Most
people
are
still
winning
with
Mike
but
anyway.
Yes,
we
can
do
14
for
111
I'm
15
for
112
I.
Guess
try
to
remember
that
I
did
the
fixes
in
some
I
think
I
did
the
fixes
in
that
city
manager,
because
so,
if
you're,
if
you're
generating
per
debuff
15,
there's
a
little
bit
of
a
problem,
I'm
surprised
it
broke
for
cops,
but
we
can
have
a
look
yeah,
probably.
B
It's
you're
saying
it
only
were
a
break,
some,
it
won't
it
which
which
command
breaks.
When
you
generate
the
protobufs,
see.
A
A
Client
I
think
was
the
case
on
that
CD
manager
like
a
different
version
of
like
gr,
PC
or
protobuf,
or
whatever
it
was,
and
so
that
that's
the
issue
that
you
sort
of
had
this
like
inclusive
versioning,
going
on
and
so
I
think
boy
didn't
seem
that
d-mat
overlay
because
he
stopped
generating
part
of
us
using
it
introduction
but
yeah
it
was
expedient,
I!
Think
yes,
so
that's
it's
not
it's
not
a
big
thing
that
that's
that's.
A
B
A
A
So
that's
been
a
sort
of
hard
blocker.
If
we
can
get
around
that,
then
we
can
switch
to
using
basil
I'd
like
to
do
that.
Because
then
we
have
reproducible
builds
and
it's
much
much
faster
and
it
works
better
in
a
CI
environment.
And
all
that's
what's
that,
because
we
don't
have
to
use
docker
to
build
things
and
docker
and
it
worked
less
well
on
a
CI
environment.
Then,
like
the
nice
basil
rules
anyway
for
now,
I
think
make
it
should
be.
D
A
Yeah
I
think
we
all
basically
yeah,
which
is
which
is
annoying
because
we
would
like
to
have
the
have
the
version
that
we
actually
shipped
be
the
one
that
we
all
develop
with,
but
there
has
been
a
blocker.
So
that's
why
I
sent
the
slightly
distasteful
PR
to
like
rewrite
the
the
weave
vendor
code
to
be
basil
compatible
so
that
we
can
get
we
can
get
to
basil
land?
A
H
H
Friend,
okay,
on
the
the
last
meeting
which
I
missed.
Unfortunately,
the
the
aspect
of
the
new
AWS
instance
types
and
the
ephemeral
stores
and
whatnot
came
up
and
I
was
wondering
what
the
recommended
workaround
right
now
for
mounting
them,
and
possibly
using
them
for
farlap
docker
or
the
couplet
is,
if
you
guys
have
any
recommendation.
I.
A
H
Only
have
one
minute
left
so
so
last
question
on
that
does
node
up
actually
do
something
or
register
itself
as
a
system
D
unit,
because
I
could
try
to
use
the
hooks
and
and
before,
but
I
don't
see
that
system
D
unit
on
a
completely
bootstrap
cluster
node.
A
B
Yep
I've
I'm,
not
sure
I've
done
some
things
with
hooks.
Where
you
know
you
don't
stall,
even
if
it
runs
too
late,
I
install
it
and
then
I
restart
the
node,
which
sounds
a
little
funky,
but
it
that's
how
we
do
like
the
app
armor
hook
and
things
like
that.
So
there's
some
there's
some
ways
to
make
it
work.
Basically,
you
can
run
any
arbitrary
staying
in
a
container.
So,
okay.