►
From YouTube: CNCF Harbor's Community Zoom Meeting - Jan 11, 2023
Description
CNCF Harbor's Community Zoom Meeting
A
All
right,
hello,
everyone
today
is
the
January
11th
and
that's
the
official
community
meeting
for
Harbor
as
Harbor
is
one
of
the
cncf
Projects.
Please
follow
the
call
of
conduct.
My
name
is
earlene
vasilev
and
I'm.
The
community
manager
for
Harbor
with
that
said,
I'm
gonna
I'm,
sorry
send
the
link
to
the
community
meeting
notes
into
the
chat.
So
please
add
yourself,
add
your
topics
and
we
can
kick
off
with
this
one
I'm
gonna
share
my
browser,
so
we
can
show
that
thing
and
I
can
see.
B
B
The
initial
timeline
is
from
January
3rd
to
April
17
in
this
year,
and
the
proposed
the
wire
Milestone,
like,
like
the
page,
showed
a
little
a
little
different
compiled
with
previous
release
planning.
We
add
a
new
milestone,
name,
the
feature
free.
What
I
mean
is
we're
giving
a
time
frame
for
the
community
and
the
maintenance
to
either
the
proposed
items
or
features
in
this
release.
So
we
have
a
Time
window
to
discuss.
J
features
or
items
could
be
include
in
this
release.
B
So
during
that
time
we
can
discuss
whether
this
feature
can
be
done
in
the
studies
or
not
if
any
features
or
atoms
there
is
a
contributor
from
Community
would
like
to
add.
They
also
can
discuss
with
the
maintainer
and
to
contribute
on
that.
So
that
is
a
little
difference.
Compare
with
previous
release.
So
that's
something
we
would
like
to
propose
in
this
release.
B
B
B
B
Yeah,
the
first
one
is
adopt
the
old
size
spec.
We
want
that
1.1.0.
This
is
something
we
heard.
We
have
not
noticed
from
the
outside
team.
They
may
have
a
plan
to
release
the
new
spec
around
the
February
or
March
in
this
year.
So
I
think
that
is
very
important
for
Harbor
as
a
registered
project
to
support
the
new
spec
as
early
as
possible.
B
We
we
still
have
a
few
items
you
need
to
complete
in
this
release
and
the
third
one
is
a
we.
We
traded,
we
traded
as
a
P1
atom
Cloud.
Even
integration
I
think
cloud
unit
is
a
either
functionality.
Chat
have
been
provided
by
many
census
projects,
but
Hubble
country
do
not
have
this
integration,
so
we
propose
to
do
some
integration
with
Cloud
event
in
this
release.
B
So
this
is
a
major
element
we
proposed
and
the
legs
the
items
we
list
the
initial
2008
item
here
that
we
propose
to
do
and
the
last
line
about
the
candidate
list.
If
I
think
only
women
not
to
go
through
each
of
these
in
this
currently
so
can
you
open
the
candidate
list?
Yeah
just
just
show
the
community
what
we
are
doing,
the,
how
we're
doing
the
plan-
and
we
will
add
some
label
like
candidate
2.8.0
in
some
GitHub
issue.
B
D
B
A
B
B
Yes,
that's
something
we
trying
to
do
in
this
release.
So,
okay,
the
idea
is,
give
you
a
time
frame
for
maintenance
community
to
include
or
remove
some
items
from
the
initial
planning
before
the
future
phrase,
based
on
what
the
resources
we
have
and
based
on
the
priority
and
based
on
the
I
mean
the
investigation.
Progress
on
that
to
do
the
to
do
the
adjustment
for
the
plan
before
the
future
free.
A
Great,
don't
get
me
wrong,
don't
get
me
wrong
if
we
can
release
during
coupon
or
on
the
first
day
of
kubecon.
That
would
be
fantastic
because
we'll
be
able
to
promote
that
at
kubecon,
and
we
can
add
that
to
our
maintenance
track.
Guys
give
me
a
second
someone
is
looking
at
my
door.
I'll
be
I'll,
be
back
in
a
second.
B
B
Think
there
may
be
still
some
risks,
because,
based
on
the
experience
for
the
previous
release,
we
may
have
some
I
mean
delayed
due
to
some
CV
need
to
fix
right
because
the
CV,
you
cannot
control
it.
It
would
come
in
when
we
release
before
we
release
that.
So
probably
we
can
reach
the
geodate.
I
mean
the
initial
targeted.
So.
D
B
E
B
E
And
as
I
know,
some
a
client
team
like
another
cosine
on
Niners
are
working
on
the
101
adoption
and
they
are
doing
some
changes
to
support
the
latest
update
of
upstream
and
development,
and
they
want
to
be
the
first
year
to
support
the
latest
chain
from
oci.
So
can
work
best.
Somehow
some
of
the
client
team
are
working
on
the
adoption,
so
the
from
the
perspective
of
register
team.
So
we
need
to
first
to
support
the
notice
oci
advertise
so
and
based
on
the
release
date.
A
C
E
A
Okay,
is
it
possible
like
to
lay
down
the
foundations
for
the
adoption
and
then
maybe
finishing
up
on
like
on
I,
don't
know
on
first
batch
release
or
something
which
is
not
good
at
all
is,
but
if
we
want
to
make
it
like
official
that
we're
gonna
support
it
in
Q8,
maybe
that's
that's
an
option
to
bend
a
bit
the
rules
If
he
if
he
wanted
to
have
it
I'm,
not
sure.
What's
what's
the
effort
behind
this
one
at
all,
but
I'm
just
yeah.
B
Just
speculating
yeah
warning
I
think.
Actually
you
you
point
out
a
good
point.
Think
about
that.
If,
if
the
old
says
spec
only
only
can
can
be
released
on
schedule
like
February,
so
if
the,
if
they
walk,
if
there
is
a
lot
of
work,
you
need
to
adopt
that.
Maybe
we
can
delay
a
few
weeks
for
today
to
complete
that,
but
in
case
they
also
I
wonder
One
cannot
be
released
in
February.
Maybe
it
would
it
it.
It
also
created
to
March
or
April
something
like
that.
B
F
Okay-
and
we
have
to
think
that
the
1.1
spec
is
already
specified-
it's
not
yet
released.
This
means
that
the
probability
that
something
will
significantly
change
apart
from
documentation
in
in
1.1
will
not
be
that
big,
I
guess
or
am
I
wrong,
so
that
the
probability
that
it
will
change
in
a
way
that
impacts
the
implementation
is,
is
very
low
because
I
think
the
mostly
it's
like
the
documentation
that
will
change.
B
F
C
D
A
Maybe
we
can
get
in
touch
with
the
LCI
folks
and
and
see
how,
if
they
see
that
if
they
can
say
that
they'll
change,
much
or
I,
don't
know
somehow
to
to
check
with
them
about
the
specification
and
if
they
they
foreseen
some
big
changes.
But
as
I
said,
if,
if
there
are
other
projects
that
are
already
working
on
this,
one,
most
probably
won't
be
much
of
a
change.
A
E
E
Yeah
but
but
but
they
cannot
guarantee
anything
yeah
so
yeah.
They
told
me
that
there
are
still
some
materials
are
under
discussing
within
their
group,
but
they
told
me
that
they
won't
be
back
change
from
RC
to
GA,
but
it's.
This
is
not
an
official
message.
Just
a
personal
information,
so
I
mean
that
nothing
can
be
currently.
So
that's
why
I'm
asking
for
several
weeks
of
time
buffer
for
the
coding?
Oh.
A
D
A
Think
if
we
know
if
we
know
that
there
is
a
possible
risk
as
I'm
changing
it,
if
they
delay
too
much
the
official
release
and
and
we
will
not
be
able
to
implement
it,
then
we
can
address
it
at
this
time,
I
think,
but
until
then
I
I
personally
I,
don't
see
a
reason.
Why
not
to
work
on
this
one
Jolene?
What
do
you
think
you
have
some.
B
Yeah
yeah
I
think
this
is
a
high
priority.
We
need
to
work
on
this,
but
what
do
we
just
talk
about
in
case
in
case
there
are?
Some
risk
is
like
the
delay.
The
release
I
mean
also
have
related
release,
and
there
are
a
lot
of
big
change.
We
need
to
adopt
and
need
a
lot
of
efforts
from
our
set
to
change
it,
to
adopt
the
new
spec,
so
that
probably
will
delay
the
release
right.
That's
something
we
just
now
talk
about.
B
A
B
C
A
C
B
B
C
B
A
A
C
F
Yeah
somehow
it
got
it
slipped
and
it
appears
now.
You
know
yeah
hunting
Us
in
this
Helm
chart,
but
the
problem
is
more
fundamentally
in
the
job
scheduler.
So
we
have
a
job
scheduler
and
if
we're
running
on
kubernetes,
we
cannot
have
multiple
instances
of
the
job
scheduler
because
of
the
fact
that
the
job
scheduler
is
mounting
a
volume
and
that's
a
problem.
E
E
The
bank
I
I
mean
a
lot
of
that
to
us,
because
that
the
the
this
volume
is
just
for
the
CVS
Port
to
support
some
country
files.
It's
not
going
to
do
the
amount,
it
should
be
just
to
do
the
do
it
in
the
temporary
folder
inside
the
part
or
container.
So
when
you
yeah
yeah,
maybe
you
Club
the
effects
for
this
problem
and
and
remove
the
amount
of
value
amount
and
just
use
the
internal
folder
yeah
go
ahead.
F
Yeah
I
think
this
will
not
fix
the
arrow,
because
I
mean
it
will
fix
the
error,
but
the
problem
will
still
be
working,
because
if
you,
if
you
swap
the
volume
Mount
to
Temporary
volume-
and
you
have
like
two-
three
four
pots-
you
know
like
for
job
service
and-
and
you
want
to
get
the
cve
every
time-
you
do
a
requests.
Every
fourth
request
will
fail
because
it
will
not
find
the
file
because
the
file
is
directed
to
a
job
service.
F
E
C
E
F
F
A
I
have
a
few
stuff.
First,
one
I've
sent
a
link
to
a
shared
document
that
we
have
to
write
up
the
proposal
for
the
maintenance
track.
So
if
you
can
take
a
look
at
this
and
if
you're
willing
to
to
talk
on
the
maintenance
truck,
add
yourself
add
your
topic
that
you
want
to
cover
there,
so
we
can
submit
that
until
27th
of
January
maintenance
track
and
not
automatically
approved
this
this
time
they
go
through
the
same
process
as
any
other
cfps,
so
we
have
to
craft
it.
B
Yeah
I
think
maybe
for
detail
content.
We
can
corporate
offline
on
this
topic
on
this
quick
comment
in
the
check
session,
but
maybe
we
we
can
also
discuss
the
speakers
this
time
for
this
session
right.
We
so
only
and
the
team
you
will
be
go
in
person,
right,
yep,
okay,
so
I
think
from
Beijing
team.
We
we
currently
still
working
on.
C
C
A
B
E
A
D
B
A
I
think
we
should
work.
We
can
work
on
this
one
asynchronously
and
if
we
can
finish
it
up
until
you
go
out
on
your
New
Year's
holidays.
So
that's
next
Friday
will
be
great
because
I
have
to
submit
it
until
the
the
Friday
after
so,
if
you
can
fix
it
until
next
Friday,
it
would
be
great
okay.
Thank
you
all
right,
I'll
add
my
topics
about
Community,
which
can,
if
we
have
enough
of
other
topics,
technical
topics
that
we
want
to
do
on
the
maintenance
track.
A
We
can
easily
remove
the
community
stuff,
but
in
in
that,
in
that
maintenance
track,
I
think
we
should
include
the
other
subproducts
projects
like
the
operator
and
the
updates
there
and
the
and
the
adoption
of
the
terraform
provider
as
well,
because
there
are
also
two
major
things
that
we
are
under
the
harbor
organization
these
days,
so
yeah
they're
in
the
in
the
same
pot.
So
we
should
talk
about
that
as
well.
A
A
Great
next
thing
are
folks
from
VMware
asked
me
about
this:
show
it's
called
the
enlightening
with
Kelly,
so
I'm
gonna
I'm
gonna
be
guest
there
next
week.
I
think
I'm,
sorry,
not
Kelly
Whitney.
So
if
you,
if
you
want
something
specific
to
to
mention
on
that,
that
show
dropping
a
line
because
it's
gonna
be
like
educational
about.
A
However,
what's
Harbor
how
Harbor
Works
architecture
bloody
blah,
but
if
you
want
something
specific
that
you
think
that
you
should
be
definitely
there
just
drop
your
line,
so
I
can
build
up
my
agenda.
It's
practically
a
non-agenda
show
it's
more
like
a
natural
conversation
driven,
but
still
I
want
to
have
agenda.
Of
course,
so
I
don't
get
lost
and
I
cannot
all
the
things
that
you
you
want
to
have
there.
So
we
can
use
that
as
a
as
a
showcase.
D
A
A
Yeah,
you
can
check
the
the
other
things
that
that
Whitney
did
and
actually
last
time
it
was
lee
Chappelle
and
it
was
about
about
Carvel.
So
you
can
see
the
the
whole
thing
it
was
before
that
it
was
open
Telemetry.
She
touches
a
lot
of
topics
and,
and
it's
a
like
a
free
chat.
What
the
thing
is,
how
it's
used?
It's
pretty
cool.
A
E
A
So
yeah
I'll
definitely
touch
base
on
this
one
that
we
will
be
working
this
one
on
the
next
release
and
I'll
mention
that
also
the
release
date
will
be
around
kubecon
and
stuff
like
this,
so
yeah,
okay
in
the
last
and
very
interesting
topic,
it's
about
the
church
Museum
deprecation
there
was
a.
There
was
a
interesting
questions,
the
other
day
about
Detroit
Museum
and
the
application.
Why
and
why
we
decided
to
deprecate.
It
I
think.
F
F
Less
a
question:
why
deprecated
it's
more
a
suggestion
that
while
we
deprecated
hosting
Helm
charts,
it
might
be
useful
to
still
support
replication
from
chart
Museum
and
also
that
you
can
import
images
into
chart.
Museum
I
know
that
you
know
you
can
convert
charts
from
chart
Museum
into
oci
quite
easily.
It's
not
no
big
deal
and
I
think
it
will
be
also
quite
easy
to
implement
a
replication
logic
where
we
fetch
images
from
chart
Museum
and
replicate
them
automatically
to
oci
inside
registry.
F
F
E
F
Be
part
of
the
replication
right
so
like
for
the
convenience
and
and
ease
of
use.
You
know
you
don't
want
to
have
a
CLI
tool
or
another
tool
that
does
this.
You
would
like
to
have
it
in
in
Harbor
that
Harvard
can
do
it
periodically
automatically
as
part
of
the
application
you
know,
so
this
is
my
take
on
it.
Maybe
it
would
make
sense
to
think
about
it
and
and
implemented.
F
E
Yeah
yeah
I
understand
that,
but
which
two
do
I
choose
and
why
the
that
is
required.
It
is
needed
some
customer
side.
We
thought
that
there
are
several
force
that
can
convert
Trump
identify
into
the
LCI
charge
and
which
one
do
we
choose,
and
so
that's
why
I
I
personally
believe
that
we
can
provide
a
tool
outside
of
the
scope
to
help
users
to
convert
that
old
and
old
style
files
into
the
oci
and
compatible
files.
G
F
But
you
have
to
take
into
account
like
if,
if
people
are
using
like
charts
in
chart
Museum
format
in
oci,
they
can
definitely
convert
their
own
charts.
You
know-
and
this
is
but
this
is
not
the
target
group.
The
target
group
are
people
who
are
using
Harbor
in
a
way
where
Harbor
is
a
replication
machine,
so
they
use
Hardware
in
a
way
that
Aggregates
images
and
charts
from
other
registries
like
Docker
Hub,
like
artifact,
Hub
and
and
so
on.
So
they
replicate
all
those
third-party
images
internally
into
their
Hardware
instance
and
then
use
it.
F
So
they
cannot
I
mean
they
can,
of
course
convert
those
third-party
images.
You
know,
but
they
have
to
write
a
tool
for
that.
You
know
and
if
it
would
be
part
of
Harbor
replication,
it
would
be
really
easier
for
those
for
those
people
to
be
somewhat
backwards,
compatible
or
or
not
backwards
compatible,
but
you
know
be
able
to
replicate.
Also
the
Legacy
chart
Museum
format,
because
the
thing
is
that
chart
Museum
will
not
go
away.
F
It
will
always
be
there,
so
it
is
not
it's
not
it's
not
oci
is
not
the
Next
Generation
storage
for
Helm
charts,
it's
one
of
other
ways,
so
the
chat
Museum
will
still
be
there,
and
people
would
continue
using
chart
Museum,
you
know
so
it's
you
know.
The
chat,
Museum
or
oci
is
not
the
next
generation
of
Helm
charts.
It's
one
storage
options
that
Helm
chats,
provide
and
I
think
if
we
are
able
to
replicate
charts
from
chart
Museum
into
oci,
and
we
don't
have
to
use
a
chat
museum
for
that.
F
You
know
we
can
implement
it
ourselves,
but
the
question
is,
you
know:
should
we
do
it
and
I
would
argue
that
would
you
know
benefit
a
lot
of
users
of
Harbor?
If
we
do
this
and
they're
going
to
be
thankful
if
we
implement
this
functionality
for
them
and
they
won't
be
that
set
in
when
we
deprecate
Chuck
Museum,
because
I
think
this
will
impact
a
lot
of
users
when
we
depreciate
chat,
Museum.
B
Yeah,
actually
we
have.
We
have
the
GitHub
issue
to
check
and
remove
the
challenging
2.8,
based
on
the
previous
editions
that
the
community
have
made,
given
that
all
our
wedding
just
said
that
replication
to
support
the
television
to
osai
artifact,
but
another
question
is
I
want
to
show
out
is
the
verification
functionalities
that
are
supporting
the
Hubble
is
using
the
sum
of
the
common
API
right
to
to
the
artifact
reputation.
B
Personally,
I
heard
from
some
of
the
Nintendo
we
are
trying
to
do
something
like
to
make
the
replication
either
spec
to
the
community.
If
we
have
the
some
specific
function
to
China
Museum
I
mean
convert
China
Museum
to
a
osai,
compatible
artifacts
Shadow.
We
are
not
a
common
cases
for
the
replication
ESL
right.
C
E
But
if
we
decide
to
replicate
the
charm
Museum-
and
this
should
be
a
specific
scenario-
not
not
different
from
other
supporting
adapters,
we
have
to
call
the
Trap
Museum
API,
rather
than
the
the
standard
way
to
a
distribution
API
to
to
manifest
or
or
just
download
files
and
then
and
push
into
the
harbor
side.
I
I
I
I
still
personally
believe
that
Richard
did
not
do
this
in
a
herbicide
because
they
speak
the
standard
and
adapter
standard.
E
You
know
Harborside,
we
always
call
the
V2
I
mean
the
distribution
API
to
cool
and
push
manifest,
instead
of
called
some
specific
server
API
to
download
files
to
do
the
reputation
so
yeah.
But
we
can
discuss
sorry.
We
can
initialize
some
proposal
in
the
community
science.
Sometimes
we
can
discuss
in
the
program.
H
Hey
yeah:
is
it
specific
to
this
particular
release
or
can
I
have
one
question
you
can
ask
sure
yeah
actually
you're
from
here
here
or
so
we
are
using
Harbor
from
the
initial
days
onwards,
we
haven't
enabled
any
life
cycle
policy
on
the
report
levels
So.
Currently
we
got
pile
up
with
terabytes
of
images
in
the
S3.
So,
even
though,
if
we
are
running
the
JC
collection
garbage
collection,
it's
unable
to
delete
the
terabytes
of
images.
So
is
there
any
way
we
can
selectively
delete
the
images
this.
H
H
F
I
think
the
problem
here
is
that
the
garbage
collection
is
not
properly
working,
in
my
opinion,
on
S3
buckets.
So
this
is
my
personal
opinion
here,
but
so
far
as
I
as
I
see
it
also
internally.
If
the
repository
is
a
few
terabyte
of
size,
we
have
no
repository
that,
where
the
garbage
collection
is
working
correctly,.
E
H
D
G
H
D
D
E
Yeah
distribution
will
Loop
all
the
tags
and
remove
the
mobile
one,
but
it
will
take
a
a
lot
of
times,
but
the
default
timeout
from
distribution
to
S3
is
the
one
minute
I
believe.
So
in
some
cases
it
will
come
out,
so
you
can
upgrade
to
the
2006
and
harbor,
introduce
the
affiliate
hardware
and
to
skip
this
video
and
continue
to
remove
other
candidates.
G
D
E
Yeah
it's
in
our
background.
We
noticed
that.
But
but
since
the
garbage
crash
is
now
blocking
operation
and
I
mean
it's,
it's
not
block
any
two
approach,
operation
of
your
users,
so
we
that's
why
we
choose
the
single
server
to
to
remove
all
the
candidates.
But
yes,
this
this
will
case
the
the
performance
problem.
E
D
H
C
B
C
H
D
F
Chat
museum
is
not
the
binary
is
not
very
stable
it
and
the
other
solution
would
be
implemented
ourselves.
You
know
like
implementing
track
Museum
API
in
in
Harper,
parallel
to
oci,
so
I
think
the
reason
was
to
to
deprecate
it
in
favor,
because
there
is
already
a
solution
for
that,
but
the
reality
is
that
or
Helm
did
not
decide
to
to
switch.
You
know
from
the
old
chart,
Museum
format
to
the
oci,
only
but
they're
running
side
by
side
you
know,
and
so
yeah.