►
From YouTube: CNCF Harbor's Community Zoom Meeting - Nov 30, 2022
Description
CNCF Harbor's Community Zoom Meeting
Harbor 2.7 release information...
Proposal for new way to engage community
Discussing the Harbor Satellite/Edge proposal..
etc..
A
Okay,
hello,
everyone,
my
name
is
Julian
vasilif
and
I'm.
The
community
manager
for
Harbor
today
is
November
30th
and
that's
the
official
meeting
for
for
Harbor
project.
So
please
follow
the
code
of
conduct
and
just
be
nice
to
each
other.
I
have
issues
with
my
camera
today,
so
I'm.
Sorry
for
that
cannot
share
my
my
video
but
I'm
gonna
share
my
screen.
A
C
A
A
A
A
C
B
I've
been
thinking
about
this,
since
our
discussion
on
Monday
and
I
was,
you,
know,
stepped
back
or
stepped
one
I
did
one
step
back
and
thought
about:
okay,
how
how
we,
you
know,
get
feedback
from
the
community
or
how
we
accept
PRS
in
Harbor
from
the
community
based
on
what
criteria.
So
the
current
suggestion
is
to
do
it
based
on
voting
or
suggestion
if
it
is
desired
from
the
community
and
then
we
added
or
accept
it
or
not.
This
is
one
option
how
to
do
it.
B
That
was
a
principle
based,
so
we,
if
we
would
have
coming
out
if
we
were,
if,
if
we
have
a
direction
in
Harvard
where
Harbor
is,
is
moving
to
what
does
Harbor
support
functionality
wise.
So
then,
depending
on
this
principles
and
criteria,
every
contribution
that
fits
into
this
criteria
is
automatically
accepted.
So
we
don't
have
to
ask
the
feedback
from
the
user,
because
the
desire
needs
to
be
there.
B
If
someone
takes
an
effort
and
invests
a
lot
of
time,
you
know
weeks,
maybe
even
months,
to
create
a
PR,
because
you
know
for
us:
it's
maybe
a
few
days
to
note
coding
and
then,
but
for
someone
who
is
new
to
to
Harbor
it's
it's
like
10
times
more
time.
You
know
to
implement
something.
So
someone
has
a
really
really
big
need
to
implement
it
and
and
make
it
PR
and
I.
B
Think
if,
if
we
can
base
our
assumption
about
what
is
needed
in
in
Harbor
or
not
based
on
the
principles
that
we
have,
then
we
don't
have
even
to
ask
the
community
if
they
accept
it
or
not.
So,
for
example,
if
we
say
we
ex,
we
support
the
top
top
four
Cloud
providers.
You
know
like
we
want
to
make
sure
that
Harbor
is
running
on
the
the
top
top
four
Cloud
providers
and
if
someone
is
suggesting
you
know,
extensions
or
Helm.
Chart
updates
for
Helm
chart
update
that
in
both
Azure.
B
This
is
in
line
with
our
principles
that
we
want
to
support
it.
Then
it's
clear
that
we
should
accept
this.
Pr
same
goes,
if
I
say
in
a
hypothetical
way,
we
shouldn't
do
it,
but
in
a
hypothetical
way
we
want
to
say
that
we
would
like
to
move
Harbor
as
a
general
purpose,
artifact
repository,
and
then
it's
totally
fine,
if
someone
accepts
you
know,
proposes
a
PR
and
we
accept
it
that
supports
Maven.
B
You
know
you
know,
so
you
get
the
picture,
what
I
mean
here
and
if
we
would
have
this
principle
or
if
we
would
communicate
and
make
those
principles,
public
and
I
think
we
have
a
common
understanding
in
the
community
where
we
want
to
see
Harbor
in
the
future
and
where
we
want
to
position
it.
But
if
we
would
write
it
down
and
make
it
explicit,
then
we
wouldn't
have
to
ask
the
community
if
we
should
accept
this
PR
or
not,
because
then
it
would
create.
Does
it
fit
in
our
principles
in
our
strategy?
B
If
yes
send
it
in?
If
not,
then
not
you
know,
and
so
this
is
this
is
I'd,
say
a
suggestion
how
we
can
see
this
this
thing
as
well,
because
I
think
getting
the
involvement
from
the
community
getting
a
voting.
A
decent
voting
from
the
communities
would
be
a
bit
difficult
and
then
we
need
to
decide
at
what
stage
do
we
accept
a
vote?
You
know,
should
it
be
three
votes
or
should
it
be
30
votes?
And
you
know
we
need
to
discuss
this
topic
as
well?
Yeah,
okay.
B
A
A
A
B
A
A
And
for
me
that
label
or
PRC
is
in
particular
to
be
able
to
select
the
PRS
that
needs
more
attention
from
broader
audience
in
in
terms
of
usage.
Let's
put
it
this
way,
because,
in
the
end
of
the
day,
the
the
main
core
maintenance
team
is
taking
the
decision
for
specific
features
in
PRS
if
they
should
be
accepted
or
not
right.
So
it's
like
a
semi-democracy.
C
A
B
A
And
for
me
that
thing,
if
we
want
to
use
it
and
to
say
the
community,
a
that's
the
process
that
we
want
to
ask
you
for
your
feedback
on
something
or
do
you
share
something
with
us?
That
should
be
like
easy
to
swallow
kind
of
words.
So
it's
should
I,
don't
know
how
to
put
it
right,
but
it's
like
super
welcoming
and
actually
but
on
the
other
hand,
to
be
like
yeah.
We
definitely
need
your
input
here.
So
please
do
that.
A
A
C
D
D
E
A
And-
and
that's
like
that
particular
voting
for
that
issue-
is
the
exact
example
or
I
need
such
kind
of
approach.
Let's
see,
and
then
the
next
thing
yeah
right,
defining
the
label
is
just
super
simple
right,
but
then
we
have
to
work
with
the
community
to
educate
them
to
to
follow
that
mindset
like
yeah
like
we
can
do
some
kind
of
automation
or
something
that
just
send
to
the
mailing
list
or
whatever
a
list
of
issues
or
PR's
that
are
labeled
like
this.
So
we
can.
A
A
A
A
D
D
A
D
D
A
D
A
B
They
do
the
same
thing
right
so
I
think
makes
sense
to
to
to
to
you
know,
break
the
community
together
here
and
I'm
working
on
that.
Maybe
we
should
sync
sync
on
that
one
or
the
other
day
and
I
can
tell
you
where
we
currently
stand
and
what's
what's
the
the
idea
behind
it
there
and
I
would
like
to
the
next.
The
next
community
meeting
I'm
gonna,
do
a
presentation
about
that.
E
So
right
now
we
are
creating
an
open
source
completely
and
we're
planning
to
contribute
it
at
some
point
right
now,
because
we
had
problems
with
getting
things
merged
Upstream.
We
just
created
our
own
open
source
repo.
All
the
code
is
already
open
source
and
fine
with
contributing
it.
It's
just
a
cube
Builder
operator,
but
that
also
is
like
another
point.
We
had
some
problems
getting
things
merched
in
Harbor
operator,
because.
E
A
E
C
C
E
A
C
A
E
B
B
A
Okay,
but
of
course,
when
we
speak
in
the
in
terms
in
long
term,
like
relation
with
the
project
and
maintaining
and
everything
that
that
shouldn't
be
on
the
table
for
doubts,
I
think
I
I
can
I
can
talk
normally
correct
me,
but
Steven
Steven
zoo
is
the
the
one.
That's
normally
running
this
kind
of
effort,
yeah.
D
D
But
actually
he's
not
working
on
operator
I
mean
Harbor
this
program
and
yeah
I
I
I,
actually
I
I
had
a
meeting
with
the
engineers
in
the
operator
and
maintenance
group
and
last
week
and
Lee
may
not
take
much
effort
to
to
help
us
to
maintain
the
operators.
So
the
the
decision
is
that
I
and
change
yeah
will
spend
some
time
in
future
to
contribute
to
the
harbor
elevator
to
help
them
to
review
PR
text
file.
Obviously
the
operator
to
align
with
the
hyperosis.
So
the
current
a
situation
is.
D
D
D
A
D
C
C
E
And
we're
right
now
in
a
phase
where
we
want
to
spend
a
lot
of
time
on
this,
and
we
also
want
to
spend
some
time
on
what
we
discussed
previously
in
operator.
That
also
manages
the
harbor
front
end,
and
if
we
can
more
easily
integrate
that
in
Harbor
operator,
instead
of
creating
like
a
separate
operator
project,
then
I'm
also
happy.
C
B
Okay,
cool
I'm
I'm
also
working
on
a
concept
here
and
to
get
some
consolidation
in
this
space
because
there
are
like
the
terraform
provider,
there
are
some
other
providers
and
we
would
like
to
I
would
like
to
consolidate
it
and
see.
There's
a
big
demand
here
and
I
think
we
should.
You
know
and
there's
other
people
working
on
it.
I
know
it.
E
The
first
use
case
is
to
use
harbor
as
a
registry
cache,
and
for
that
we
need
to
deploy
and
configure
Hardware
through
git
Ops.
We
can
deploy
it
through
gitlops
with
the
operator
already,
but
managing
and
setting
up
whichever
application
strategy.
We
want
it's
not
possible
through
githubs
right
now
and
we
would
run
it
on
like
60
plus
clusters,
where
we
don't
want
to
do
that
manually.
C
A
D
A
E
A
A
C
A
B
I'm,
actually
not
so
again,
30
minutes,
okay,
I
mean
we
can
start
with
the
RC.
You
know
rc2
or
C1
by
that
day
and
then
just
flip.
It
flip
the
switch
on
the
third
or
fourth
of
January.
With
the
new
release
we
can
do
the
same
or
we
can
release
before.
Christmas
I
mean
it's
nice
Christmas
present
I'm,
you
know.
B
C
E
E
C
B
Yes,
we
can
okay.
So
what
is
Harbor
satellite?
The
satellite
concept
has
been
discussed
in
Harbors
for
quite
some
time,
and
we
are
now
at
the
stage
where
we
would
like
to
implement
it.
We've
been
discussing
this
concept
with
a
few
customers
before
and,
and
we
think
very
strongly
that
there
is
a
need
for
the
solution.
So
what
is
have
our
satellite
Harbor
satellite
is
a
container
registry
that's
running
on
the
edge
specifically,
the
container
registry
is
running
on
the
edge
cluster
itself.
It's
a
registry
that
is
stripped
down.
B
B
The
the
other
use
cases
Global
image
distribution.
So
you
would
like
distribute
your
images
around
the
globe
and
all
your
Edge
cluster,
and
then
you
would
flip
a
switch
and
all
image
applications
will
be
updated
immediately
and
without
the
delay.
So
you
would
not
stress
your
main
infrastructure
with
image
pulls.
B
So
now
the.
How
does
it
work
and
how
does
it
look
like?
So
we
have,
you
know,
created
this
concept,
which
called
satellite
and
Ground
Control.
You
know
satellite
is
out
in
space.
You
know
flying
around
the
globe
and
the
ground.
Control
is
the
thing
that
controls
it
and
it
gets
commands
or
it
sends
commands
it
receives
data
and
the
same
concept.
We
would
like
to
apply
with
the
harbor,
so
we
would
like
to
extend
Harbor
with
a
new
API
and
new
functionality,
which
is
called
Ground,
Control
and
Ground.
B
Control
is
by
itself
just
managing
the
satellites.
The
things
like
image,
replications
or
copying,
images
from
the
from
them
from
from
from
Main
Harbor
to
the
satellite,
is
done
via
the
existing
mechanisms
in
Hardware.
So
we
have
already
the
tools
to
do.
You
know
pro-based
image,
replication
or
push-based
image
replication
to
satellites
and,
as
you
can
see,
we
have
in
Harbor
and
the
functionality
which
is
ground
control
and
Ground.
Control
is
a
receiving
entity,
so
it
receives
API
calls,
you
know
a
connectivity
so
like
you
can
register
a
satellite
to
produce
app
control.
B
B
So
how
does
satellite
look
like
satellite
is?
We
would
really
really
specifically
or
have
kubernetes
first
approach
or
why
is
it
changing?
We
have
a
kubernetes
first
approach,
meaning
that
we
design
a
satellite
primarily
for
kubernetes.
It
might
be
running
as
a
binary
itself
outside
of
kubernetes,
but
this
is
not
our
primary
goal.
Primary
goal
is
having
a
satellite
image,
so
it's
basically
one
application.
It's
a
one
binary
that
contains
the
core,
so
I
call
it
core,
because
it's
analogous
to
Harbor
core.
B
It's
not
the
same
thing
as
the
harbor
core,
but
has
quite
a
few
similarities.
You
know.
Maybe
we
can
copy
a
few
things
over,
maybe
not
I,
don't
know,
then
we
have
the
registry,
the
registry.
Is
we
either
take
the
registry,
the
docker
distribution,
that
is
there
or
we
select
another
one?
It's
currently
discussion,
which
registry
we're
going
to
choose,
it's
not
yet
decided.
B
So
then
we
will
package
it
all
in
one
in
one
binary
which
will
contain
an
sqlite
so
sqlite,
because
it's
light
and
we
don't
have
to
store
that
many
data
and
the
only
requirement
that
Hardware
satellite
has
is
a
persistent
volume
where
images
are
stored
and
the
database
is
stored.
So
this
is
the
only
requirement
that
we
have
on
the
satellite
core.
Satellite
core
has
also
the
admission
controller.
It's
an
optional
feature
that
you
can
enable
disable
in
your
in
your
cluster
and
the
admission
controller
does
the
following.
B
So
if
you
want
like
the
use
case
is
the
following.
So
if
you
would
like
to
deploy
your
application
on
the
edge
you
don't
wanna
or
the
developer
should
not
be
thinking
about
all
right,
which
registry
should
I
do
in
this
deployment.
So
what
the
developer
does
is
just
deploys
the
application
to
Cluster
a
as
to
Cluster
b
as
to
Cluster
C
in
the
same
way.
So
the
key
has
the
same
URL
inside
the
cluster
inside
the
same
URL
or
same
image.
B
B
So
it
will
not
point
to
an
external
registry,
but
it
will
always
point
to
the
satellite
and
you
can
configure
how
this
rewrites
should
happen
so
that
we
write
all
pods
best
specs
to
satellite
core,
so
that
if
you
have
a
proxy
and
things
like
this,
it
will
you
know
kick
in
or
if
you
want
to
just
rewrite
the
registry,
that
your
own
registry
and
not
nothing
else.
So
you
just
if
someone
is
pulling
from
Docker
Hub,
it
will
still
work
you
know
or
using
Docker
app
as
a
reference.
B
It
will
still
work
so
that
mission
controller
is
capable
of
doing
a
rewrite
and
is
doing
the
rewrite
for
the
for
the
for
the
Pod
specs,
so
that
the
user,
the
developer,
who
deploys
an
application
to
Edge,
doesn't
have
to
deal
with
Hardware
satellite
in
any
way,
so
it's
transparent
and
automatically
rewrite.
So
you
create
rules
where
you
can
deploy
rules
from
from
control
to
to
the
satellite,
and
this
is
happening
in
in
Via,
Ground
Control
or
maybe
you
don't
have
any
rules
at
all.
B
You
don't
have
the
option
to
control
that
much
rules
just
enable
position,
control,
yes
or
no,
and
the
admission
controller
is
the
one
who
is.
You
know
listening
to
events
from
from
the
kubernetes
API
and
then
knows
if
something
is
deployed
to
the
to
the
cluster
and
then
can
rewrite
it.
So
this
is
the
basic
concepts
that
we
try
to
implement.
B
So
our
request
from
the
community
is
feedback.
You
know
we
would
like
to
have
feedback
from
the
community
about
the
use
cases
about
their
scenarios,
that
we
can
think
and
incorporate
them
into
the
into
the
design
and
development.
We
of
course
interested
in
community
contributions
in
form
of
implementation
and
code
contributions,
reviews
testing
and
so
on
and
from
the
hardware
maintainers.
We
would
like
to
get
feedback
regarding
the
feasibility
so
which
components
can
be
used.
What
things
can
we
reuse
and
what
can
we
swap
out
and
yeah?
B
B
D
B
D
D
D
B
In
addition,
Ground
Control
is
not
responsible
for
image
transfer,
so
images
are
not
transferred
via
Ground
Control.
The
images
are
transferred
via
normal
flow,
that
is,
it
is
already
already
there
like
a
replication,
so
we
want
to
build
upon
the
existing
features
of
of
Harvard
so
that,
like
replication
and
yeah
is
used
to
distribute
images
to
satellites
and
Ground.
Control
is
just
accepting
commands
and
you
know
sends
data
to
the
to
the
satellite.
B
B
D
B
B
Yeah,
so
when
you,
when
you
deploy
a
satellite,
the
satellite
needs
to
connect
to
ground
control,
and
this
connectivity
process
or
onboarding
process
of
a
satellite
needs
to
happen.
Somehow,
and
you
need
to
provide
a
URL
to
your
ground
control
and
you
need
to
provide
an
you
know
a
secret
or
onboarding
secret
or
something
so
that
the
the
satellite
can
register
itself
into
Ground
Control.
And
then
you
can
see
all
your
satellites
in
Harbor
and
then
control
those
satellites.
D
D
D
B
B
This
is
something
that
it's
typical
in
the
kubernetes
cluster,
so
you
have
namespaces
and
you
want
assign
permissions
per
namespace.
This
is
I,
think
the
typical
concept
that
we
would
like
to
implement,
and
so
you
don't
have
a
really
complex
permission
set.
There
is
no
users,
you
just
say
what
proof
secrets
you
want
to
inject
in
which
namespace,
so
the
the
authorization
and
authentication
Concepts
here
are
much
simpler.
There
is
no
option
to
push
images
to
to
satellite
core,
pushing
pushing
images.
B
Yeah,
it's
out
of
scope,
yeah
and
that's
why
I
I'm
not
sure
if
we
can
or
how
much
you
can
take
over
from
from
Harbor.
But
the
my
goal
is
not
to
Fork
Harbor
core
and
you
know,
take
everything
out
or
something,
but
to
see
that
we
can
kind
of
a
more
Sherry
pick
functionality
that
are
well
well
designed
in
Harbor
that
we
can
take
over.
D
B
B
B
E
Yeah,
so
you
have
an
admission
controller
that
somehow
has
to
edit
different
pots.
I've
worked
on
admission
controls
quite
a
bit
and
I
can
be
very
complex
for
the
submission
controller.
If
you
want
to
basically
modify
pots,
my
question
would
be:
do
you
also
plan
to
then
modify
deployments
demon
sets
jobs
and
so
on,
because
those
spawn
essentially
the
pots.
B
B
D
D
B
Okay,
then
I
would
that's
that's
it
for
my
site.
So
if
you
have
feedback
and
questions
to
this,
you
can
you
know
you
can
just
ping
me
on
my
my
email
address.
I'll
just
put
it
here,
my
email
ping
me
and
then
we
can
discuss
and
maybe
I
can
help.
You
clarify
a
few
questions
and
if
you
would
like
to
contribute
conceptually
and
and
on
use
case
basis,
I'm
also
happy
to
to
get
the
feedback.
A
C
A
B
Did
you
see
the
video
from
from
Victor?
It
made
a
really
nice
video
about
Harbor.
So
if
you
didn't
see
it,
you
should
someone
posted
it
in
the
slack
channel
the
other
day,
I
think
yesterday,
it's
it's
really
nice.
One.
Managing
container
images
would
help
synap
and
other
artifacts
with
Harbor.
It's
a
really
nice
video.
It's
really
well
a.
E
C
A
A
A
Thank
you
for
reminding
me
this
one
yeah
we
have.
We
received,
finally,
the
the
mail
for
submitting
the
maintenance
track,
so
we
have
and
they
was
a
27th
of
January
or
something
so
we
have
plenty
of
time
to
to
do
that,
but
I'm
really
Keening
on
doing
it
in
person
this
time,
all
together
with
the
kiosk
like
we
did
last
time
in.
A
So
the
conversation
varies
from
technical
to
community
to
how
to
get
involved
everything
pretty
much
everything.
However.
So
if
you
want
to
join
this
effort,
please
drop
me
a
line,
so
we
know
how
many
folks
you
have
there.
It's
not
a
strict
thing,
but
we
have
it.
What
is
that
with
him
like
a
three
days
or
two
days
and
for
a
few
hours
a
day,
yeah
so
there's
plenty
of
time
to
to
visit
any
other
things.
B
A
And
and
that
kubecon
that
is
going
to
happen
in
in
Amsterdam,
it
will
be
the
biggest
in
Europe.
So
far
we
have
twice
as
more
cfps
for
for
that
kubecon.
So
cncf,
you
know,
expects
more
at
least
50
more
percent.
More
people
visiting
and
Amsterdam
is
super
more
communicative
city
than
anything
else
in
Europe
pretty
much
as
of
now.
So
there
will
be
people
flying
or
so
I
think
we
definitely
must
do
the
maintenance
track
in
person,
so
we
can
get.
We
can
have
more
attention
from
the
from
the
crowd.
A
A
Wait
you
to
contact
me
about
this
stuff
if
you
feel
like
it
I'll
add
them
into
the
meeting
minutes
so
not
to
forget
until
next
time
to
discuss
it
once
again,
all
right!
Thank
you.
Everyone
have
a
great
rest
of
the
day
or
the
evening
wherever
you
are
and
talk
to
you
in
two
weeks
and
in
two
weeks
and
then
I
suppose
the
next
one
after
two
weeks
it
will
be
somewhere
around
Christmas,
so
we'll
be
skipping
this
one,
so
yeah,
20
28th
of
December
I,
think
we'll
be
skipping.