►
From YouTube: 20200422 - Image Builder Office Hours
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
This
record,
meaning
is
being
recorded
and
it
will
be
uploaded
to
YouTube
I've
dropped
a
link
to
the
agenda
in
the
chat,
if
you
feel
so
inclined,
please
add
your
name
to
the
list
of
attendees
and
I
will
do
that
right
now
and
so
I
put
in
some
agenda
notes
here,
please
feel
free
to
add
right
now.
It's
all
everything
I
could
think
of
which
is
basically
just
a
review
of,
what's
been
going
on
and
some
interesting
outstanding
items
to
talk
about
starting
off
with
a
few
updates.
A
A
It
was
only
me
and
one
other
community
member
two
weeks
ago,
so
this
might
be
new
to
some
people,
but
I'm
working
on
having
this
image
builder
take
over
building
an
additional
or
a
secondary
OVA
that
is
needed
for
the
cluster
API
provider.
For
vSphere
project,
it's
a
H,
a
proxy
load,
balancer
oba
I'm
porting-
that
over
that
PR
should
be
open.
Today
it
helps
immensely
to
have
the
same
tool
generating
Oh
having
the
same
tool
generated
OVA.
So
you
don't
end
up
with
different
metadata
and
different
styles,
and
things
like
that.
A
It
will
significantly
decrease
some
maintenance
burden
for
myself
and
the
cap
V
maintainer
x',
nothing.
Also
on
the
vSphere
side
of
things.
We
had
a
wonderful
contribution
from
a
community
member,
adding
a
new
support
for
a
new
builder.
That's
been
in
Packer,
since
1.5.2
I
think
I've
got
the
version
right
and
that
is
the
vSphere
ISO
builder.
So
the
current
one
was
you
had
two
choices:
the
VMware
ISO
builder.
A
Yesterday,
I
think,
with
the
new
release
of
container
D
being
bumped
to
1.3
dot
for
Jason
to
Tigris,
opened
a
PR
to
read
that
and
that's
the
defaults
in
the
image
builder
for
cluster
API
images.
Now
so
that's
everything
that
I
know
of
that
has
been
the
major
items
that
have
changed.
The
last
two
weeks
with
in
building
images
for
cluster
API
before
I.
Go
on
to
discussion
topics
is:
is
there
anything
that
anybody
else
would
like
to
to
bring
up
so
I'm?
A
B
A
B
A
B
A
All
right,
I
will
just
do
so.
So
the
first
thing
I
put
in
under
discussion
topics
was
something
I
wanted
to
bring
up,
because
this
is
something
that
I'm
getting
from
requirements
on
my
employer,
which
is
you
know
the
image
builder
tool
right
now
for
clustering,
API
me
and
a
couple
other
community
members.
We
build
images
pretty
much
with
all
the
defaults
in
the
image
builder
as
it
exists
today.
Jason
de
tigress
builds
them
and
pushes,
and
let
me
clarify
that
this
is
for
cluster
API,
not
the
stuff
that
justin
has
just
in
Santa.
A
Barbara
is
usually
taken
care
of,
but
we've
build
a
Mis
and
make
those
available
to
the
community.
I
build
obas
make
them
available
to
the
community,
but
the
intent
of
the
cluster
API
builder
is
for
people
to
be
able
to
build
their
own,
whether
their
enterprise
users,
there
they
have
their
own
requirements
additional
packages
that
they
want
to
inject
we've
it's
a
little
ugly
as
Moshe
knows,
but
there's
lots
of
little
knobs,
where
you
can
inject
additional
packages
and
all
that
kind
of
stuff.
So
the
question
has
come
up
about.
A
How
do
we
validate
the
image
as
being
something
that
can
be
successfully
deployed
by
cluster
API
and
I?
Don't
mean
from
a
CI
CD
perspective.
I
mean
more
from
like
an
image
scan
perspective
like
boothy
image
and
what's
running,
can
we
validate
the
versions?
You
know
say
I
mean
it
would
be
conceivable,
for
example,
like
cluster
API
only
supports
container
D,
but
there's
enough
knobs
in
there
you
could
probably
rip
out
container
d
and
install
docker.
If
someone
did
that,
how
would
we
catch
it?
A
I
know
that's
a
far-fetched
example,
and
so
I
was
wondering
if
anybody
had
any
thoughts
on
this,
the
little
bit
that
I've
looked
into
it.
My
thought
so
far
would
be
the
AWS.
The
kappa
builder
right
now
is
using
goths
gos
s.
It
is
a
very
simple
thing,
as
part
of
its
Packer
provisioning
to
validate
actually
the
container
D
is
there
and
running,
and
not
much
else,
I
forget
what
all
it
is.
A
C
A
A
A
C
A
Yes,
I,
agree:
I,
think
you
could
put
a
lot
in
there
and
any
I'd
I
hadn't,
really
familiarized
myself
with
Goss
before
I.
Don't
know
what
the
origins
of
using
it
for
Kappa
are,
but
it
is
a
tool
for
server
validation
and
I'd,
rather
use
an
existing
tool
and
build
off
of
something
that
we
are
already
using
then
try
to
create
something
from
scratch,
which
which
would
probably
not
go
well.
A
C
A
D
A
That's
a
really
good
point:
yeah
I
mean
once
you
start
having
those
those
yeah,
because
it's
all
yeah
mo
for
Goths
about
what
to
check.
For
you
know
it's
it's
almost
kind
of
like
a
regression
thing:
you're
like
oh.
If
this
was
configured
wrong,
let's
add
it
and
make
sure
that
we
can
validate
in
the
future
Thanks.
A
A
couple
of
them
are
super
simple
I
mean
I,
think
you
had
referenced
in
the
past
that
you
might
have
some
colleagues
or
members
on
your
team
that
might
be
able
to
take
over
some
of
that,
but
like
it's
like
a
docker
change
and
a
Lintz
change
and
I,
they
look
good
to
me,
but,
like
I
said
last
time,
I
don't
feel
comfortable,
the
proving
stuff
that's
in
the
cube,
deploy
so
I
just
wanted
to
raise
them
and
bring
them
to
your
attention.
Thanks
for
coming
this
week.
Thank.
D
You
I
actually
I
care,
two
of
them
I
approved
to
merge
two
of
them,
and
so
yes,
it's
the
two
more
recent
ones
which
actually
a
little
trickier
around
like
this
abling
doctor,
was
actually
interesting
for
this
group.
So
we
proposal
is
to
disable
donker
when
we
pre-installed
on
her.
We
shouldn't
automatically
start
it
in
case
it
installs.
It
might
be
tables
rule
which
then
locks
us
in
to
either
nf
tables
or
legacy
mode.
D
In
a
way,
we
don't
want
so
interesting
why
it's
a
little
bit
more
surprising
than
you
might
imagine,
but
you
know
I
will
I
will
take
a
look
at
this
I
think
I
better
head
into
the
buster
image,
which
is
I
guess
are
these
two
remaining
ones
are
tied
into
a
poster
image
which
is
which
is
on
the
end.
So
that's
great
okay,.
C
C
D
A
C
C
C
A
C
A
Yeah
because
I
think
one
of
the
things
that
I
was
picturing
was
you
know
and
like
said
I,
don't
know
how
much
of
the
ansible
scripts
would
would
remain
or
not,
but
hoping
that
we
could
have
like
a
configurable
back-end
where
you
could
switch
between
hey.
You
know
I've
invested
heavily
in
ansible,
even
if
it's
not
long-term
viable.
It
exists
today
and
switch
between
it
and
your
config
atom
type
provisioner,
because
you've
invested
heavily
in
that
and
but
I
think.
The
first
attempt
at
the
CLI
was
trying
to
have
that
configurable.
C
So
the
rough
idea
of
the
the
CLI
is
to
have
multiple
phases
and
have
an
image
go
through
those
phases.
So
you
feed
an
image
in
and
then
a
face
can
be
add
something
on
top
of
that
image
test
that
image
or
convert
that
image
from
one
format
to
another.
So
and
an
image
could
be
an
ami
and
I
saw
a
row
disc
image
or
any
one
of
those
things
and
then,
based
on
what
tooling
we
have
available
to
convert
from
which
formats
to
which
formats.
You
then
have
this
matrix.
C
That
says,
if
I
start
with
a
docker
image,
acting
ready
to
roll
and
if
I
have
an
ami
I
can
go
to
a
my
Ernie
and
if
I
have
an
image,
I
can
go
to
a
my
GTR
and
whatever
it
is
so
that
you
have
this
pluggable
way
of
dealing
with
images
so
that
we
can
single
saw
something
and
they
push
it.
Where
we
want
and
then
for
those
who
don't
want
a
single
source,
then
they
can
use
their
preference.
A
B
A
C
The
the
container-based
old
is
basically
it's
just
in
a
different
image
format
that
we
support,
so
your
image
formats
would
be
container
disk
image.
A
my
GCE
and
the
job
and
I
can
go
from
a
container
to
a
disk
image
and
I
can
go
from
a
disk
image
to
AWS
sure
in
GC
and
because
I
can
do
that.
I
can
start
with
a
dr.
container
and
then
put
it
through
to
transformations
and
then
I
can
put
it
wherever
I
want
okay.
C
So
it's
just
one
transformation
step
in
front
of
starting
with
a
real
disk
image,
so
it
fits
quite
nicely
in
it.
You
just
yeah,
you
just
have
an
additional
transformation
step
in
in
the
beginning,
and
then
then
the
testing
phase
will
then
say.
Your
testing
phase
is
going
to
run
on
an
image
as
well,
so
give
it
give
me
container
container.
C
A
So
I
think,
if
you're
hoping
to
have
something
mirja
belen
a
couple
of
weeks
that
would
be
like.
Let
me
know
when
I
should
look
at
it
again,
so
I
can
kind
of
see
where
it's
going
and
yep
I
would
very
much
like
to
be
able
to
to
it
or
iterate
on
it.
I
know:
you've
got
kind
of
the
AWS
initial
implementation,
obviously
I'm
interested
in
working
on
the
vSphere
implementation.
A
A
A
A
Okay,
all
right
all
right.
Well,
if
there's
nothing
else
to
say
on
that
either
kind
of
main
interesting
thing
as
come
in
recently
as
yeah
Justin
work
on
a
docker
or
container
based
builder
I,
don't
know
if
I
just
wanted
to
put
it
on
there
in
case
there
is
something
you
wanted
to
say
about
it
Justin
or
we
wanted
to
look
at
it
together,
I've
scrolled
through
it,
but
it's
currently
out
of
my
area
of
knowledge
like
I,
looked
at
it
I
could
see
what
it's
doing
and
it
kind
of
fit
in.
D
D
D
From
that
perspective,
it
has
some
mash
scripting,
the
the
my
father
most
complicated
weather,
to
complicated
scripts,
one
of
which
is
how
do
you
wrap
a
docker
image
which
is
basically
to
our
file
and
convert
that
to
an
image
like
an
a
disk
image,
a
bootable
disk
image,
including
copying
the
kernel
and
the
bootloader
into
the
whatever
the
program
of
the
boot
must
have
boot
record
is
these
days
and
then
the
other
complicated
scripts.
D
It
is
like
getting
a
raw
disk
image
into
AWS
is
surprisingly
difficult,
and
so,
like
just
there's
a
lot
of
there's
a
really
boring
script
that
brings
up
a
temporary,
am
free
Amazon
instance
and
scps
I.
Believe
the
disk
image
across
writes
it
20
vs
volumes,
natural
EBS
volume,
others,
but
that's
just
so
Machinery.
The
core
of
it,
though,
is
like
these
two
docker
files,
I
think
they're
called
Buster
base
and
Buster
AWS
and
Buster
a
the
best
basis,
or
these
from
a
Buster
base
and
they're
very
simple.
D
Think
I
really
like
what
Moshe
said
around
the
idea
of
transformations
and
I
think
that
hopefully
muffing
Bosch
will
go
away
and
we
can
make
the
particulars
too
complicated
by
scripts,
and
we
can
essentially
get
it
to
the
point
where
you
know
the
logic
to
upload
an
image
to
upload
a
roll
disk
image
to
AWS
is
the
same,
however,
being
produced
that
image.
So
there's
no
need
for
it
to
be
specific.
D
To
this
and
I
think
being
able
to
plug
in
this
work
into
the
CLI
would
create
a
really
I
think
it
mazing
suite,
because
then
you
could
say
well,
we
use
concur
and
they
want
to
do
this
and
I
want
to
use
docker
or
to
build
the
images
when
use
basil
to
build
the
images.
I
think
that
would
be
nice,
and
then
we
have
like
tooling
that
consistently
enables
you
to
equip
them
today
to
be
us
to
gz2
the
sphere
wherever
your.
Whatever
your
target
is,
that's
me
seems
really
exciting.
So
just.
D
Am
very
I
am
very
happy
to
be
proved
wrong
here.
There
there
are
three
passes
I'm
aware
of
the
one
I
took
here,
which
is
you
basically
upload
and
snapshots
the
path
where
you
upload
to
s3
and
there's
a
imports,
image
API
and
a
third
path
which
is
there
is
an
explicit
service
called
the
image
import
service
I
tried
to
frustrate
the
s3
one
and
it
looked
like
it
only
produced
a
or
read
about
the
s3
one
it
looked
like
it
produced.
D
What's
called
an
in
store,
backed
image,
so
it's
a
different
type
of
as
opposed
to
no
EVs
type.
In
it's
a
different
type
of
image,
maybe
I
don't
have
had
to
convert
it
using
it.
Some
using
the
same
process
that
affected
me
did
anyway,
okay
and
then
I
tried
the
the
image
import
service
or
the
import/export
service.
D
A
D
C
D
D
It's
not
so
it's
not
it's
a
non-event,
it's
the
restriction
on
AWS.
Then
it
gets
only
a
restriction
on
their
image.
Import/Export,
service
and
I
think
the
I
think
every
provider
has
their
own
way
of
doing
things.
I
think
if
you
scroll
down
I,
think
there
is
still
a
script
for
GCP
import
summer
and
the
GCP
import
is
you
do
just
upload
it
and
then
you
upload
it
to
GCS,
and
then
you
make
them
cool
of
some
sort.
D
B
C
D
So
it's
not
actually,
you
know
hooked
up
to
the
bootloader
and
then
there's
a
secondary
step,
but
we
were
also
in
so
grumpy
see
again
in
the
container,
but
it
doesn't
actually
saw
in
the
bootloader
because
there
is
no
bootloader
in
a
smart
file
and
then
that's
the
sort
of
complicated
bit
in
in
the
messy
script
which
expands
the
tar
file,
builds
a
disk
image
expands
by
announced
it
expands
the
tar
file
into
a
disk
image
and
then
executes
the
grub
install
into
the
bootloader.
So
that's
a
little
bit
more
complicated
like
to
follow
along.
D
D
A
Let's
see
so
one
question
I
would
have
is
basically
next
steps
for
this
like
what?
What
do
you
need?
You
know,
it's,
it's
label
is
a
work
in
progress.
What
do
you
need?
You
know
what
could
I
do
for
testing
if
you're
looking
for
help
or
and
a
second
question
I
have
with
that,
is
like
it's
about
kind
of
repo
organization.
A
D
D
A
D
Like
that
was
a
clear
like,
let's
put
it
in
a
straw,
man
that
we
can
all
agree
is
wrong.
I
could
go
under
cube,
deploy,
Buster
or
something
like
that
or
whatever
it
is
I.
Remember
exactly
and
then
I
would
hope
we
could
use
some
Moshe
CLI
to
replace
the
scripts
in
here.
I
would
hope
we
I
don't
know
how
important
it
is
actually
to
share
the
darker
files
themselves,
but
it
would
be
nice
if
we
could
I,
don't
know
if
that
really
mmmm
docker
files
aren't
great
at
structuring.
D
D
Yes,
I
think
I
think
getting
it
going
or
getting
into
an
area
that
we're
happy
with
to
execute
in
that
direction,
and
it
would
be
good
I,
don't
know
if
you
saw
there
is
a
cloud
build
file,
so
this
can
be
built
unattended,
which
is
sort
of
nice
I
and.
D
D
That
could
work
I
just
try
to
think
the
I
think
we
still
need
some
privileges
in
terms
of
so
running,
a
doing
a
bind
amount
and
not
sure
you
can
do
the
blind
mount
or
you
can
do
the
moment
in
darker,
but
you
need
privileged
mode
execution.
So
I
don't
know
if
we
can
do
prowl
there.
So
I'm
just
wondering
whether
we
can
get
both
the
8
of
us,
Crenshaw's
and
privilege
most
privileged
mode
doctor
in
the
same
thing
but
I
imagine
we
can
figure
out
a
sequence.
We
could
always
upload.
E
D
D
C
D
C
D
Does
and
that's
so
the
cube
deploy
builder
also
depends
on
those
and
has
docs
on
how
to
create
them.
It's
because
I
think
you
basically
need
to
create
some
EPC
and
security
group.
Regardless
I
mean
SSH
access
to
it,
so
I
wasn't
sure
whether
to
pre,
where
they
try
to
create
it
on
the
fly
or
whether
they
just
assumed
that
it
had
been
created.