►
From YouTube: CNCF Harbor's Community Zoom Meeting - July 28, 2021
Description
CNCF Harbor's Community Zoom Meeting
A
As
you
can
see,
we
don't
have
a
strict
agenda
for
today,
so
it
can
be
open
discussion.
I'm
gonna.
I
have
one
topic
that
I
want
to
share.
A
B
I
think
alex
alex
said
he
was
on
the
call
he's
gonna
join
shortly.
A
A
So
alex,
do
you
have
something
to
talk
about.
C
I
don't
know
if
childhood
is
around,
but
volume
is
here
and
yeah.
I
was
looking
at
the
2.4
items.
I
think
we'll
have
a
more
meaningful
update
for
the
next
one,
we'll
have
something
to
demo
for
the
the
tracing
work
that
we're
doing
with
open
telemetry
as
well
as
the
prometheus
stuff.
C
I
think
you
know
it's
just
right
now
what
a
point
in
the
release
where
we
just
got
finished
planning
with
with
everything
we're
going
to
do
for
2.4,
so
take
a
look
at
the
2.4
project
board
that
should
reflect,
I
think
at
least
the
majority
of
the
issues
that
will
be
in
the
2.4
release,
and
then
we
need
to
further
break
these
down
into.
A
C
But
those
are
the
epics
or
the
themes
hey
no
longer.
I
wonder
if
you
can
talk
about
the
stuff
that
we're
doing
for
arm,
and
you
know
the
work
that
we've
been
doing
with
a
lot
of,
because
I
think
I've
mentioned
a
couple
times,
but
ultimately
we
weren't
able
to
get
anything
released
right.
There
was
nothing.
C
There
was
nothing
released
in
the
2.3
slipped
into
2.4
and
maybe
for
the
folks
that
were
interested
because
I
have
seen
some
tickets
in
the
past
on
you
know.
When
are
you
going
to
have
an
arm
build
of
harbor?
What
is
their
thinking
here
and
is
it
going
to
be
released
under
the
harper
repo,
the
harper
project,
or
is
it
going
to
be
under
some
other
project?
And
how
are
we
thinking
about
this.
D
Actually,
we
do
not
have
any
a
lot
of
update
about
the
arm
because
of
the
the
contributors
are
busy
on
their
company's
business
strategy.
So
we
just
escaped
the
recent
several
weeks:
okay,
but
the
the
basically
we
we
wanna
to
shape
her
arm
align
with
the
2.4
release
and
the
contributors
will
prioritize
their
their
tasks
in
the
last
months.
C
D
We
have
already
created
a
sub
project
in
the
go
harbor
river,
which.
A
C
All
right,
yeah
sounds
good.
Thanks
for
the
update.
I
haven't
been
involved
in
this,
so
much
so.
For
those
that
don't
know
it's
there's
a
company
called
aladda,
which
is
a
cloud
cloud
company
in
china.
That's
been
driving
this
with
us
previously,
but
like
going
inside,
you
know
they've
been
busy
with
their
own
work
for
some
time.
C
Thanks
dana
thanks
for
linking
the
repository.
C
Yeah,
so
I
mean
this
is
much
more
of
a
status
update
for
now
join
us
and
orlan,
there's
nothing
to
demo
for
this
one,
but
I
think
for
next
one
we'll
have
something
planned.
I
think
I
think
we'll
go
with
the
proposal
at
the
very
least
for
how
we're
thinking
about
adding
tracing
right.
So
you
can
look
at
a
request,
as
it
makes
its
way
through
the
various
components
of
harper
and.
D
C
We're
thinking
about
this
mostly
for
I
mean
identifying
bottlenecks
in
harvard
today
and
just
making
harper
run
faster
and
better,
and
we
have
you
know
we
have
some
real
performance
needs
around.
C
C
Right,
presumably
you
know,
the
the
edge
nodes
are
are
controlled
the
image
on
the
on
the
edge
notes,
everything
on
everything
on
the
edge
notes
right
there,
they're
pushed
down
from
a
central
harbor,
so
you've
already
done
your
your
security
analysis
or
your
image
scanning
and
you're
signing
and
and
all
the
policies
around
that
and
you're
pushing
down.
C
You
know
through
a
headless
service
through
hardware
applications,
and
so
when
you,
where
we're
really
there's
a
lot
of
performance,
performance-related
work,
that's
going
on
right
now
and
we're
thinking
about
how
to
make
our
replication
much.
You
know
more
resilient,
much
more
scalable
right
we've.
Never
we
know
we
can
configure
two
replications
between
a
heart,
for
instance,
and
and
another
registry,
but
we've
never
really
tested
the
use
case
where
you
have
a
hubspot
deployment
of
a
a
central
harbor
right,
a
central
harbor.
C
That
is
the
harvard
that
exists
today
and
then
rotating
out
to
a
series
of
series
of
harvard
lights
on
different
edge
sites.
Right
and,
if
you
think
about
this,
for
you
know,
for
retail
right
when
we
talk
about
edge,
we're
talking
about
retail,
something
like
ups
or
starbucks
or
potentially
looking
at
you
know,
anywhere
between
100,
harvard
instances
on
edge
notes
to
possibly
a
thousand
or
ten
thousand,
and
so
that's
kind
of
the
scale
that
we're
dealing
with
that's
kind
of
the
forward-looking.
C
They've
set
it
up
just
like
I've
like
I've
talked
about,
we
have
worked
with.
This
is
a
customer
of
vmware,
and
I
think
you
know
through
that
customer
we
we've
really
been
shaping
and
driving
some
of
these
requirements
around
harvard
light,
but
that's
something
we're
going
to
be
releasing
for
the
oss
community.
Like
I
said,
but
yeah
I
mean
the
basically
the
the
use
case
that
the
use
case
with
edge
right
now
is
that
the
hardware
instance
on
the
edge
site
is
not
does
not
have
to
be
full
featured
right.
C
It
does
not
have
to
have
all
the
compelling
features
that
exist
in
harvard
today,
because
there's
a
single
persona
behind
that
right,
like
all
the
the
different
personas
that
we
have
in
harvard
today
with
guests
and
developer
project
admin
system,
admin,
there's
you
know,
you
don't
need
that
many
personas
on
the
edge
site,
because
it's
operated
by
a
single
person,
and
so
that's
going
to
be
a
much
lighter,
a
much
more
stripped
down
version
of
harper
and
the
biggest
concern
or
the
biggest.
C
You
know
thing
that
we
have
to
be
careful
about
right.
Now
is
really
the
replication
right
because
with
edge
sites
they're,
not
24
7,
you
know
highly
available
right,
they
might
go
down
if
what
happens
if
you're
doing
replication
and
the
site
goes
down,
sometimes
these
retail
sites
don't
get
powered
back
up
until
maybe
you
know
hours
or
24
hours
later
and
ideally,
the
hard
replication
should
be
resilient.
D
C
That
it
just
picks
up
where
left,
where
you
left
off
right,
which
you
can't
do
right.
You
can't
do
that
right
now,
so
yeah,
the
replication
is,
is
like
the
biggest
the
biggest
improvements
that
we
need
right
now.
I
think,
and
then
you
know
we
have
to
make
harvard
smaller
for
it
to
run
on
the
edge
site
right.
Harbor
right
now
is
too
big.
It
takes
too
much
memory.
It
takes
too
much
cpu
and
we've
heard
typically
on
the
next
site.
You
know
we're
working
with
anywhere
between
a
one.
C
You
two,
you
know,
maybe
likely
a
two
you
right,
possibly
a
for
you
server
right,
just
a
single
server
or
two
servers
on
in
half
a
rack
of
hardware
space,
and
you
know
that
has
to
run
the
entire,
the
entire
kubernetes,
this
drill
right,
plus
the
hardware
registry.
So
it's
shared
with
the
rest
of
your
your
stack
right.
If
it's,
you
know,
obviously,
if
it's
a
vmware
stack,
vmware
customers,
there's
the
vmware
software
data
center.
C
But
if
you,
if
you
are
another,
if
you're
using
another,
you
know
kubernetes
distribution,
possibly
rancher
or
or
anthos
or
openshift
or
some
other
diy
cloud,
then
it's
the
same
problem
right.
You
can't
it's
not
just
running
a
harbor,
it's
running,
harbor,
plus
the
rest
of
your
kubernetes
stack
and
so
harper
as
it
exists
today
is,
is
way
too
big
for
that
and
there's
no
need
for
it
to
be.
You
know
to
take
that
much
resources,
because
a
lot
of
these
features,
I
think,
won't
won't
get
leveraged
very
heavily.
C
I
I
don't
think
customers
will
be
pushing
you
know
from
the
edge
side
to
the
central
harbor.
It's
going
to
be
one
way
for
a
lot
of
these
applications
for
a
lot
of
these
actions,
and
you
know
they
have
to
be
centrally
gated
scanned
signed
and
then
they're
pushed
out
to
the
edge
sites
so
that
that's
kind
of
the
assumptions
that
we're
working
with
so
does
that
sort
of
answer.
Your
question
yep.
C
All
right,
yeah,
that's
that's
really!
All
I
wanted
to
share.
A
I
I
wanted
to
bring
the
topic
about
the
notification
for
the
new
releases,
like
the
one
from
the
other
day
yeah.
I
really
like.
A
I
really
want
to
to
do
more
often
notifications,
so
even
for
the
minor
releases
so
can
we
can
we
work
on
some
some
kind
of
process
how
we
can
get
like
notification
internally
through
the
hardware
team
and
and
and
then
I
can
work
on
the
notification
through
the
my
necklace,
feeder
or
whatever,
because
that's
that
those
minor
releases
are
not
scheduled
really
right.
They're
on
on
demand
pretty
much.
A
C
Yep,
so
there
is
a
harvard
channel
within
our
company
slack
that
gets
notified
from
from
different
ci
runs,
but
I
don't
think
it's
limited
to
just.
I
don't
think
it's
the
action
of
uploading
it
to
github.
It's
really.
We
produce
a
successful
build,
but
that
kind
of
you
know
it's
a
precursor
to
what's
about
to
happen
right.
You
can
use
that
as
to
kind
of
know
that
hey
we're
getting
ready
to
release
a
build
and
then
hey
lion,
is
there
a
way
to
send
notification
to
slack
every
time
or
to
email?
C
Well,
yeah
I
mean
orlan
is
thinking
about.
How
does
how
can
he
get
alerted
right
because
he's
not
in
our
internal
engineering
channel
he's
in
project,
maybe
or
maybe
we
can
announce
it
in
the
project
harvard
channel
as
soon
as
as
soon
as
as
you
upload
the
container
images
or
you,
you
know
you
are
releasing
it
up.
D
Should
we,
I
can
try
to
refer
to
other
cncf
projects
to
to
know
how
their
notify
users
about
their
release.
Views
are
not
released
and
but
actually,
as
I
know,
they
just
send
a
message
myself
in
their
stack
channel
right.
C
B
Can't
we
just
have
that
in
the
release
process
as
a
manual
task
there,
when,
whenever
there's
a
release,
make
sure
that
we
announce
that
in
the
public
channel.
C
D
C
A
Okay
and
and
from
community
stand
point
of
view,
it's
more
engaging
to
show
that
you
release,
often
and
people
can
can
be
more
engaged
and
to
follow
that
so
right,
at
least
when
I,
when
I'm
using
some
software,
if
I
get
to
notify
that
new
releases
are
coming
up
and
not
to
follow
specifically
a
website
or
something
but
like
through
twitter,
engages
me
more
in
in
general.
A
So
I
think
we
can
apply
that
and
I'm
gonna,
I'm
gonna
look
if
we
can
find
some
some
way
when
we
do
the
the
release
on
github
to
hook
it
to
the
public
channel
in
slightly
and
then
we
can
kick
off
the
rest
of
the
process
of
twitter
and
stuff
manually.
A
But
until
then
I
I
think
it
would
be
great
if
we
can
update
the
process
for
my
new
notification
in
china.
Whoever
is
doing
the
release
this
time.
C
C
A
And-
and
the
other
thing
that
I
wanted
to
talk
about-
is
the
oh
to
discuss
a
way,
how
we
can
get
more
organizations
to
contribute
in
terms
of
to
kind
of
distribute
the
work
not
to
be
so
vmware
centric,
because
I
was
doing
that
analysis
of
the
elephant
factor
for
harvard
and
it's
currently
it's
one,
which
means
the
90
of
work
is
done
by
some
somehow
represented
by
vmware
employees,
which,
personally,
as
a
community.
A
C
How
did
you,
how?
How
are
you
getting
this
metric
of
one,
because
I
know
there
are
the
ovh
guys
have
still
been
pretty
active
with
the
operator?
They
didn't
make
consistent
contributions,
but
how
is
it.
A
Getting
measured
the
measurement,
I
did
it
through
be
honest
with
you
correct
me
about
the
name,
but
what's
inside
the
byte.
A
Yeah
all
right
and
yeah
by
getting
the
the
data
out
of
github
and
calculating
the
the
elephant
factor.
So
it's
very
simple
calculation.
But
for
now
it's
so
it's
a
ratio
between
total
commits
and
and
commits
made
by
the
the
first
one
or
two
company
organizations
and
vmware
is
far
more
exceeding
a
number
than
anyone
else.
Combined.
A
So,
which
is
it's
not
necessarily
a
bad
thing,
but
in
terms
of
drawing
the
project
outside
of
the
only
vmware
influence,
I
think
we
should.
We
should
put
some
effort
there
and
that's
in
my
task
list
for
the
next
month.
A
If
I
can
attract
so,
I'm
gonna
practically
approach,
some
some
some
more
organizations
and
so
try
to
convince
them
to
be
more
active
and
to
put
some
more
effort
in
in
contributing
in
different
ways,
not
only
by
tools
but
documentation
and
and
more
test
cases
and
more
like
a
practical
examples
of
implementing
harbor.
A
C
You
know,
they're
not
consistent
enough
right,
and
this
is
not
to
blame
anyone,
but
you
know.
I
think
this
has
really
been
something
that
we've
struggled
with.
We
try
to
be
mindful
of
everyone's
responsibilities
within
their
own
companies.
Right,
not
everyone
is
getting
compensated
to
to
actually
work
on
harper
but
yeah.
I
feel
like
you
know
it
comes
and
goes
there
were.
There
were
releases
over
the
last
two
years
that
you
know
they
were
filled
with
external
contributions.
C
At
times,
then
there
were
other
releases
where
it
was
very
hard
to
get
something
going
right
and
we,
you
know,
I
think
we
we
have
certain
dates,
that
we've
committed
to
right
and-
and
you
know
that
forces
us
to
sort
of
you
know,
draw
a
cutoff
date
for
when
features
can
get
kick
and
get
merged
because
it
impacts
our
it
impacts
our
ability
to
release
on
time
right.
We
always
try
to
release.
C
C
You
know
work
that
we
wanted
to
to
release
in
the
harbor
and
in
the
harbor
in
a
particular
harbor
release
right
and
they
get
pushed
out
and
sometimes
they
never
get
completed.
And
I
just
wonder
you
know
how
do
we,
like?
You
said:
how
do
we
get
first
of
all
get
more
contributors
to
come
to
the
table
and
then
also?
How
do
we
work
with
them?
C
A
little
better
to
make
sure
that,
what's
being
you
know,
discussed,
is
actually
getting
implemented
and
follow
through
to
the
very
end
right
and
without
sacrificing
code
quality
without
sacrificing
their
own.
You
know
responsibilities
on
other
projects,
so
it's
just
you
know
just
being
a.
D
C
A
balance,
but
also
getting
more
contributions
I
have
I
will.
I
will
say
that
in
the
last
two
releases
it
was
extremely
vmware
heavy
yeah.
C
D
C
Get
something
when
they
get
something
else
within
their
own
company,
then.
Obviously,
that's
the
utmost
importance
in
in
the
harbor.
The
os
stuff,
unfortunately
suffers.
I
think
you
know
it's
it
all.
It
obviously
works
best
if
they're
working
on
something
that
depends
on
their
work
on
harbor
or
is
tied
to
their
work
in
harvard
in
a
big
way
right,
the
stuff
that
they
work
on
for
their
own,
their
employers,
and
so
I
think
you
know
this
isn't.
C
This
is
not
to
be
disparaging
anyway,
but
I
think
I
think
one
of
the
ways
we
can
do
this
is
just
sort
of
lay
out
the
you
know
the
goals
with
what
we're
trying
to
do
with
these
work
streams
with
external
contributors
right.
C
And
just
be
very
clear
on,
you
know
what
they
need
and
why
they
need
it
ultimately,
and
how
does
that
impact
their
product?
You
know
some
of
their
integrations
and
things
like
that
I
found
you
know.
I
found
those
kind
of
partnerships
to
be
the
most
fruitful
is
is
when
you
know
exactly
what
everyone
is
coming
to
you.
C
You
know
what
they're
coming
to
you
for
and
why
they
need
this
feature
to
be
implemented
in
a
very
certain
way.
B
Alex
you're
you're
touching
on
something
that
tianon
wrote
here
in
the
chat.
Where
are
those
things
communicated?
We
need
to
update
the
roadmap?
B
Yes,
it's
really
out
of
date
right
now,
so
it's
it's
our
commitment
to
the
community
when
we
publish
the
roadmap,
so
everyone
knows
what
we're
working
on
and
that
makes
it
easier
for
people
to
contribute
as
well,
because
otherwise,
if
they
don't
know
what
we're
working
on,
they
don't
know
if
they
can
contribute
so
having
the
roadmap,
updated,
linking
out
to
proper
tracking
issues
for
new
features
and
or
milestones
or
epics.
Or
what
have
you?
It's
super
important.
C
C
You
know
the
major
work
streams
in
a
particular
release
and
then
and
then
you
know,
we
use
the
project
board,
but
we
also
use
the
milestones.
The
milestone
label
on
the
issue
is
actually
actually
tells
you
exactly.
C
You
know
if
it's
being
worked
on
in
the
next
two
weeks
right,
if
it's
an
active
development,
that's
that's
really
it.
I
don't
think
we
have
anything
else
that
that
addresses
what
what
tienen
is
asking
for
yeah
and
I
think
there
I
need
to.
C
I
need
to
write
like
a
goals
and
or
philosophy
document,
that
kind
of
highlights
where
the
project
is
going,
because
the
project
I
feel
like
has
taken
off
pretty
well,
but
yeah
there
are
other
there
there.
You
know
some
major
themes
in
the
future,
as
we
think
about
expanding
into
you
know
harvard
as
a
service
harbor
for
edge.
You
know
other
just
other
distributions
of
hardware
like
like
like
harbor
arm,
and
things
like
that.
C
I
think
it's
gonna,
be
you
know
all
the
feature
work
that
we're
doing
are
still
really
important,
but
I
feel
like
that's
almost
just
one
thing
right.
That's
just
that's
just
you
know
future
work
for
harvard,
and
there
are
all
these
other
larger
themes
here
that
we
need
to
want
to
lay
out
for.
C
C
A
C
The
sprint
is
just
two
weeks:
it's
not
possible
to
see,
I
I
don't
think
it's
possible
to
see
which
sprints
are
in
a
particular
release,
but
you
know
those
are
roughly
the
dates
for
the
sprints,
so
you
can
see
what's
being
worked
on.
You
know
in
the
next
two
weeks
or
next
four
weeks.
Basically,
and
the
project
board
is
really
for
you
know,
issues
it's
it's
for
you
know.
D
D
A
C
There's
no
particular
order
between
these
two.
It's
just
the
project
board
is
used
to
kind
of
get
a
view
of.
You
know
the
major
features
planned
for
release
and
then
the
roadmap
or
things
we'll
be
doing.
C
A
And
about
this
one
about
the
edge,
do
we
have
it
like
a
requirement
from
specific
organization
that
they
want
to
have
it
and
we
just
start
working
on
it
or.
D
C
C
So
I'm
talking
about
vmware
in
this
case
they
have
a
real
need
around
a
hybrid
light
for
edge
deployments
and
that's
kind
of
where
we've
taken
our
cues
and
been
you
know,
hashing
out
the
requirements
but
yeah.
I'm
I'm
happy
to
chat
about
this
on
the
community
call
as
well.
I
have
thought
about
this
before,
but
not
you
know,
do
it
to
a
really
with
a
really
detailed
proposal,
because
it's
in
flux
in
her
for
downstream
as
well.
It's
just
you
know.
I.
D
A
A
I'm
asking
because,
as
far
as
I
know,
for
example,
companies
like
ericsson,
they
deploy
a
lot
of
stuff
in
in
their
edge,
and
I
know
that
we
have
people
from
eriksen
engaged
with.
If
I'm,
if
I'm
not
mistaken.
So
maybe
we
can
just
an
idea.
We
can
approach
them
to
collaborate
on
this
one
because
inevitably
they'll
need
that
functionality
as
well,
if
they're
aiming
towards
harvard
to
be
in
production
use
for
their
batch
solution.
For
example,.
C
E
E
A
E
Have
we
have
multiple,
multiple
use
cases
where
we're
using
hardware
edge?
Actually,
two
customers
yeah
at
the
moment,
they're
using
or
did
the
building
solution
already
so
that
they're
not
using
yeah
they're
building
already
something
themselves.
E
Yeah
I
mean
there
would
be
both
of
the
customers
would
be
happy
to
contribute
to
harvard.
So
it's
not
it's
not
something
they
do
on
secret.
A
E
But
we
need
to
know
the
the
requirement
and
the
scope
currently
because
I
asked
for
a
yeah.
We
need
to
know
what
is
the
status
like
the
proposal
where
we
can
contribute
on
first
on
the
document
basis
and
then
see:
okay,
what
what
other
customers
have
and
how
can
this
be
adapted,
what
they
have
and
what
they're
working
on,
because
it's
not
yet
ready,
and
basically
they
mostly
is
one-
is
using
multiple
hardware
instances
full-blown
one
and
the
other
one
has
a
more
customized
version
yeah.
A
That's
great
and
alex
how
we
can
drive
that
one
I
mean
what
is
the
so,
maybe
I'm
asking
stupid
questions,
but
how
how
it's
working
to
to
set
up
this
kind
of
discussion
and
and
and
start
like,
as
being
said,
contributing
to
the
the
proposal
and
the
actual
idea
of
the
implementation.
C
C
So
I
can
yeah,
let's
chat
offline
and
let's,
let's
get.
D
C
A
C
Yeah,
it's
a
cap
or
it's
a
series
of
caps
but
yeah.
We
we
pushed
a
proposal
to
the
community
repo
and
we
don't
have
a
process
to
really.
You
know
for
people
to
vote
on
it.
I
think
we
just
internally,
you
know,
get
consensus,
and
then
you
know
what
we're
fairly
good,
with
making
sure
everyone's
comfortable
and
aligned
right
and
then
we
just
start
working
on
it.
C
A
Yeah
I'll
create
a
pool
in
the
in
the
slack
channel
who
wants
to
participate
in
that
discussion,
and
then
we
can
chat
with
you
how
to
set
up
the
working
group
and
the
proposal
document
and
then
kick
it.