►
From YouTube: Carvel Community Meeting - June 16, 2022
Description
Carvel Community Meeting - June 16, 2022
We meet every Thursday at 10:30am PT. We'd love for you to join us live!
This week we went over the latest releases in kapp-controller and secretgen-controller, what the team is working on this week, and discuss packagerepositories. See full agenda here: https://hackmd.io/F7g3RT2hR3OcIh-Iznk2hw?both#June-16-2022-Agenda
A
A
That's
Thursday
and
it's
at
10
30
Pacific
time,
and
this
is
like
an
opportunity
for
you
to
come
in
and
just
talk
with
the
maintainers
and
see
what
we're
working
on
and
if
you
have
any
questions
or
anything
just
if
you
want
to
just
feel
free
to
pop
in
we
are
a
very
nice
Community.
We
really
like
talking
to
people.
A
That
says,
discussion,
topics
and
just
add
there
any
kind
of
discussion
topic
that
you
want
to
to
get
us
to
talk
about,
or
you
have
any
questions
or
anything
we'll
just
jump
there
and
try
to
get
everything
so
when
you're
joining
us,
please
I
just
ask
you
to
add
your
names
here
so
that
we
know
who
was
present
and
yeah.
So
that's
that
so
that's
get
started
with
our
meeting.
So
let's
go
start
into
our
announcements.
So
this
week
the
we
do
have
some
blog
post.
That's
coming
out.
A
B
A
Let
me
check
for
like
the
next
meeting,
do
we
have
anything?
So
this
is
the
week
of
the
June
16th?
Oh,
we
do
have
so
apparently
next
week
we're
going
to
have
another
blog
post
from
one
of
our
colleagues
Kumari
tanushri,
it's
going
to
be
about
targeting
the
Practical
scenarios.
Example
update
deployments
when
config
map
is
updated,
so
we'll
add
that
to
our
community
meeting.
A
A
If
you
be
creative
and
we're
waiting
for
your
contributions
and
if
you
are
not
a
Carl
maintainer,
we
have
a
surprise
for
you,
because
if
you
contribute
with
a
blog
post
or
something
something
on
those
lines,
we
have
a
t-shirt
that
we'll
be
more
than
happy
to
send
it
to
you.
So
that's
it.
Let's
move
on
to
our
releases
the
any
product
to
release
this
week.
A
D
C
I
think
we
did
mention
already
some
of
the
stuff
last
meeting,
but
just
small
recap
is
secret.
Gen
controller
introduced
a
secret
template
API
that's
available
and
you
know
give
us
any
suggestions
on
enhancements.
For
that
API
and
for
cap
controller
we
did
introduce
Joe
actually
worked
on
it,
a
feature
called
package
lading,
that's
been
also
documented,
so
give
it
a
try
or
let
us
know
if
there's
any
questions
around
it.
D
C
That
is
true.
Yes,
we
did.
We
did
the
make
the
deployment
of
secret
gen
to
be
more
similar
to
cap
controller,
which
now
both
of
them
should
not
require
any
modification
for
openshift,
and
then
the
cap
controller
release
did
also
include
some
enhancements
that
were
putting
into
cap
CLI.
That
also
makes
things
more
openshift
friendly.
A
A
B
B
If
you
look
back
not
directing
usual
but
like
if,
if
someone
just
watch,
you
look
back
a
couple
of
weeks,
we
actually
have
a
blog
post
that
gives
you
a
quick
start
like
getting
started
using
an
experimental
feature
in
ytt,
encourage
you
to
check
that
out
and
we'll
make
the
announcement
once
we
get
that
release
out
there,
but
it's
starting
to
come
together.
Looking
forward
to
that.
A
All
right,
so
the
other
only
two,
the
only
other
two
that
has
been
worked
on
this
week.
Apparently
it
has
been
the
image
package
so
who
wants
to
talk
about
image
package?
I
can
do
it
wrong?
Thank
you.
So
what
basically
we've
been
doing?
On
the
last
week,
we've
been
focusing
a
little
bit
on
performance
because
we
rolled
out
the
command
called
The
Scribe.
A
That
shows
you
all
the
images
that
are
associated
with
a
particular
bundle,
but
it
takes
a
lot
of
time
for
it
to
churn
through
all
the
images
and
try
to
find
what
images,
what
they
are
and
try
to
go
through,
like
one
lobbying
bundled
bundle,
so
we're
trying
to
look
if
there's
like
a
way
we
can
make
it
quicker
if
there's
like
any
path
on
the
code.
So
that's
basically
what
we've
been
working
on
on
on
the
last
week,
yeah
is
there
like
any
other
tool
that
have
any
about?
A
D
B
D
A
D
You
know
I
put
that
in
there
I
have
this
co-worker
who's
been
spending
a
lot
of
time,
thinking
about
package
repositories
and
I
I
thought.
Maybe
the
community
meeting
would
be
a
good
venue
to
have
that
discussion,
but
I
wanted
to.
You
know,
check
the
room's
temperature.
Maybe
you
know
not.
Everybody
cares
about
package
repositories
in
the
same
way,
but
maybe
that's
why
it's
an
interesting
topic.
A
If
you
bump
that
product
one
version
up,
you
need
you
have
like
10
new
packages
that
contain
that
point
to
10
new
bundles,
which
means
that
they
become
the
very
big
bundle
that
needs
to
be
transported
between
Registries.
That
has
become
a
problem
in
in
some
scenarios
where
people,
if
they
are
like
in
an
air
gap
environment,
they
have
to
I,
don't
know,
for
example
like
copy
10
gigabytes
of
data
and
the
next
release
that
they
do
it's.
A
A
There's
there's
not
like
a
unified
way.
Some
people
do
a
little
bit
of
this,
a
little
bit
of
that
and
there's
not
like.
Let's
call
it
a
wiener
way
for
using
package
repositories,
some
people,
some
people
due
to
the
nature
of
their
products
as
well.
They
use
a
package
repository
as
a
repositories
of
these
joint
packages
where
each
person
can
install
a
package
from
that
repository.
So
if
you
think
in
something
like
an
apt
from
Ubuntu
repository
or
a
yam
repository,
it's
something
around
those
lines.
A
There
are
people
that
don't
use
them
at
all
and
just
use,
for
example,
like
they
use,
there's
a
git
repository
that
contains
definitions
of
the
packages
or
something
like
that,
or
they
just
add
the
packages
by
hand
on
the
Clusters
there's
other
people
that
create
what
what
we
have
known
to
be
called
a
meta
package
that
basically
it's
a
package
that
in
itself
installs
other
packages.
So
if
you
think
about
it,
for
example,
you
have
an
application
that
that
contains
your
code
that
contains,
for
example,
postgres
database
that
you
want
to
install.
A
You
can
create
like
a
meta
package
that
basically
creates
packaging
styles,
to
install
your
application,
plus
the
postgres
and
so
on.
So
that's
now
pattern
that
we
saw
and
there's
another
pattern
that
is
like
multiple
gigabytes
of
assets
versus
a
single
application
asset
so
like
there
are
people
that
are
Packaging
just
one
application.
So,
for
example,
like
we
talked
with
rapidmq
folks,
and
they
just
package
right
with
them
too
right
and
they
have
a
package
repository
just
for
packaging
to
wrap
them
to
you,
sir.
A
So,
like
they're,
like
different
people
that
are
using
package
repositories,
it's
like
slightly
different
ways
but
they're
all
like
they
all
work
and
for
cap
controller,
so
yeah.
So
that's
that
what
else
the
majority
of
people
are
using
an
image
package
bundle
to
support
the
assets.
So
everything
that
you
have
like
the
package.
The
package
repository
contains
like
a
folder
that
contains
the
definition
of
the
packages
and
they
they
create
a
a
an
image
package.
Bundle
with
that
and
provide
that
as
an
image
package.
A
Bundle
to
cap
controller,
even
though
cap
controller
does
allow
you
to
do
this
in
different
methods
like
you
can
provide
a
git
repository
or
just
a
regular
CI
image.
The
people
that
we
talked
to
tend
to
use
an
image
package
bundle
on
this
before
I
continue
like.
If
you
have
any
questions,
just
feel
free
to
ask
I
feel
like
I'm,
just
machine
gunning.
Everybody
here
with
a
lot
of
information,
so
feel
free
to
stop
me
at
any.
Given
point.
Okay,.
B
A
Yes,
so,
like
the
majority
of
like
these
scenarios,
people
have
an
end
goal
of
a
narrow
gaps
of
the
environment.
So
what
is
an
air-gapped
environment?
It's
a
new
environment
that
does
not
have
access
to
the
internet.
So
all
the
images
that
are
associated
with
your
applications
need
to
be
somehow
copied
through
a
thumb
drive
through
like
a
pass-through
like
a
a
machined
middle,
or
something
like
that,
so
they
need
to
copy
all
the
these
assets,
all
the
oci
images
that
are
related.
A
For
example,
let's
pick
up
like
the
rabbitmq
example
that
need
to
copy,
like
the
Revit,
mq,
OCR
images
to
the
to
to
the
internal
networks,
so
image
package
does
provide
you
with
simple
command
a
way
to
download
all
the
images
that
are
associated
with
the
package.
Repository
bundle,
it
just
downloads
everything
and
you
can
just
copy
it
to
the
side
and
just
push
it
back
to
the
other
registry
in
a
seamless
way-
and
this
is
this-
is
like
a
very
powerful
for
for
these
scenarios,
where
you
need
to
copy
a
big
amount
of
images.
B
C
You
I,
would
you
know
maybe
to
to
push
on
that
a
little
bit.
You
can
still
use
like
a
git
repository
right
to
to
to
store
a
bunch
of
packaged
CRS
and
those
package.
Crs
themselves
are
bundles.
Let's
say
that
somebody
made
right,
they
don't
have
to
be,
but
let's
say
they
are,
and
those
things
could
be
relocated
right
as
well.
C
It's
useful
to
capture
the
asset,
just
kind
of
a
holistically
right,
and
so
that's
I,
would
say
the
benefit
of
putting
stuff
into
an
image
package.
Bundle
where
you're,
not
you
no
longer,
are
depending
on
a
git
repository
right.
There
is
no,
you
know
you
don't
have
to
necessarily
figure
out.
How
do
you
provide
the
git
server
or
whatever
endpoint
externally,
to
your
users
right
you
effectively,
shifting
that
problem
to
something
a
little
bit
more?
C
Maybe
commonly
exposed
or
or
shareable
right,
which
is
the
registry
servers
right
so
if
you're
able
to,
for
example,
wrap
your
thing
in
a
bundle,
your
entire
repository
right,
then
you're
able
to
transfer
all
of
those
packages
recursively
to
somebody
right
versus.
If
you
are
maybe
yourself
consuming
those
packages,
or
maybe
you
have
a
very
tight
feedback
loop
between
teams
and
whatnot.
You
know
there
might
be
just
no
need
to
kind
of
return.
The
git
repo
into
a
more
shippable
or
transferable
asset
right,
yeah,
yeah.
B
C
Don't
know
git
server
or
something
like
that
right.
So
you
could
be
doing
all
that
stuff
with
you
know
sure,
there's
other
ways
of
transferring
Assets
in
similar
ways
right
so
kind
of
becomes
maybe
a
preferential
kind
of
a
thing,
but
there
is
definitely
the
recursive
nature
of
which
package
bundle
definitely
makes
it
a
little
easier
to
transfer
all
of
the
dependent
child
assets
right
versus
in
git.
Then
you
would
have
to
I.
Don't
know
you
would
have
to
make
a
list
of
all
them
and
copy
them
as
well.
D
It
seems
like
a
lot
of
what
the
package
repository
story
is
air
gap.
First.
Is
that
fair,
like
it's?
Not
oh,
there's
a
central
like
if
I'm,
using
yum
or
apt,
or
something
there's
a
central
repository
right
and
I'm,
very
literally,
not
air,
gapped
I'm,
going
out
to
the
network
to
pull
my
packages
if
I'm
using
git.
D
C
There
becomes
a
question
of.
Do
you
really
want
to
do
a
lot
of
this
extra
prep
work
versus
you
know
in
a
kubernetes
ecosystem.
Registries
are
readily
available.
Production
grade
can
carry
a
lot
of
assets.
You
know,
High
availability,
et
cetera,
et
cetera,
et
cetera
versus,
like
you
know,
who's
who's
necessarily
running
a
highly
available
git
server
for
their
own
stuff,
right
or
I
mean
I.
Guess
they
could
be
using
a
managed
service
if
they
really
wanted
to.
But
it
just
seems
like
registry
is
becoming
more
and
more
of
a
service
or
API.
B
E
E
When
you
really
want
to
hide
away
the
implementation
detail
and
just
give
them
an
artifact
say:
hey
just
run
this
Commander
install
this
single
yaml
and
you
get
rabbitmq.
You
get
app
toolkit
from
tce,
you
get
whatever
it
is,
and
they
don't
need
to
care
about.
What's
underneath,
then
having
it
in
oci
bundle
kind
of
strengthens
that
approach
of
I'm,
giving
you
a
solution,
I'm,
not
giving
you
a
bunch
of
yaml
files,
I'm
giving
you
a
solution,
and
that
seems
to
also
be
a
separation
of
how
I've
been
approaching
it.
At
least.
D
B
That's
helpful
yeah
that
is
I
was
going
to
push
on
on
the
other
side
of
that
which
is
and
I
don't
know.
If
I'll
get
my
facts
straight
here.
I
do
know
that
in
most
oci
Registries
there's
some
intelligence
around
blobs
that
are
identical,
are
shared
or
can
be
shared,
but
that
in
git,
that's
not
necessarily
true
so
I
know
disc
is
cheap,
but
I
wonder
if,
like
there's
a
difference
there
in
terms
of
overtime,
a
git
repo
would
grow
much
larger
than
its
corresponding
oci
registry.
B
A
To
run
the
applications
in
kubernetes
right
and
that's
where
the
bulk
of
the
size
comes
from.
Yes,
definitely
all
right,
like
the
blobs
can
be
shared
between
between
two
images.
Blobs
can
be,
can
be
shared,
but
in
in
here,
like
the
the
smallest
part
of
any
bundle
is
or
these
package
repos
or
even
even
if
the
packages
is
a
configuration
themselves
because
they
usually
what
you
have
on
the
oci
image
that
represents.
The
bundle
is
basically
yaml,
so
in
some
sense
having
it
in
git
versus
okay.
A
That
will
change
every
time,
so
there's
no
reusage
of
of
blobs
or
having
it
in
git
in
terms
of
like
disk
space,
it
doesn't
make
a
lot
of
difference,
maybe
maybe
it
gets
smaller
and
get
because
git
only
gets
the
diffs.
While
for
for
the
oci
images,
you
need
to
you
need
it
there's
no
idea
of
leaf.
It's
always
a
new
thing.
E
I
also
always
don't
only
view
relocating
as
a
use
case
for
air
gapped
there's
a
huge
use
case
for
it
just
to
due
to
network
latency
and
things
like
that
when
pulling
an
image
instead
of
pulling
from
Docker
Hub,
where
I
may
get
rate
limited
or
even
just
pulling
from
the
internet,
if
I'm
pulling
a
lot
of
images,
I'd
rather
pull
that
from
a
local
registry.
That's
within
my
network
that
I'm
not
going
to
be
using
huge
Network
bandwidth
to
pull
down,
and
things
will
be
much
quicker.
E
Yeah
and
how
many
times
this
year
has
Docker
Hub,
GCR
or
Quay
or
GCR
gone
down,
the
answer
is
probably
multiple
times
for
everyone.
I
know:
Quay,
it's
been
like
seven
or
eight
times
this
year,
ngcr
I've
had
ghcr
I've.
Had
it
go
down
twice
on
me
this
year,
projects
registry
vmware.com
has
gone
down
on
me
multiple
times
this
year,
so
having
your
own
is
beneficial.
D
A
When
there's
like
a
new
version
of
a
particular
package,
people
really
like
to
have
the
ability
to
do
a
pull
request
or
a
merge
request,
depending
on
the
git
that
you're
using
on
the
server
side
and
and
have
like
that,
control
and
add,
like
everything
in
some
ways,
automated
we
talked
to
to
the
rabbit
like
rabbit,
mq
team.
They
had
like
a
pipeline
where
they
just
press
a
button
in
a
pipeline.
It
just
bumps
a
minor
version
of
rabbitmq
and
just
generates
yamos
for
them
and
does
pretty
interesting
stuff.
A
So
that's
that's
also
people
are
are
using.
That
was
something
that
we
found
out.
People
are
using
git
to
store
the
base
of
their
package
repository
so
like
the
sort
of
information
that
then
will
be
generated.
Another
thing
that
we
that
we
found
out
is
that
people
start
constructing
some,
some
Norms.
Let's
call
it
around.
How
can
you
define
which
versions,
which
versions
of
which
software
you
want
to
provide
on
a
particular
release
of
the
the
bundle
the
the
package
repository
they
have
like
yaml
files?
A
They
have
some
scripts
that
read
those
files
and
just
select
which
versions
they
have
because
a
lot
of
people
saving
their
git
like
multiple
versions
of
the
same
software.
So,
let's
imagine,
wrap
them
key.
You
have
like
10
versions
of
private
mq,
but
you
just
want
one
of
them
to
be
part
of
your
package
repo.
So
they
have
like
these
mechanisms,
and
we
saw
two
like
in
in
like
to
just
to
wrap
this
up.