►
From YouTube: CNCF Harbor's Community Zoom Meeting - July 14, 2021
Description
CNCF Harbor's Community Zoom Meeting
A
We're
recording
hello,
everyone,
my
name
is
olivia
and
I'm.
The
community
manager
today
is
july.
The
14th
and
is
the
meeting,
is
under
the
code
of
conduct
of
cncf,
so
welcome
everyone.
I'm
gonna
share
my
screen
with
the
agenda.
For
today
I
can
see
a
few
folks
added
topics
alex
and
daniel.
A
Yep
yep
great
so
before
the
do
that,
please
everyone
at
your
topics,
if
you
want
to
discuss
them
and
with
that
one
alex
go.
B
C
C
Sorry
I
was
talking
when
I
was
yeah
I
can.
I
can
start
first.
C
C
we've
noticed
an
issue
that
was
introduced
when
we
bump
up
the
gold
compiler
to
use
gold
one
dot.
I
think
15
to
compile
notary
there
will
you
will.
You
may
encounter
an
issue
that
when
you
try
to
sign
a
an
image
whose
repository
key
was
created
in
previous
version,
you
will
see
an
arrow
that
notary
tells
you.
C
C
Yeah,
I
think
it's
here
notary
tells
you
that
the
key
does
not
exist.
C
Right
here,
yeah
the
reason
the
root
cause
for
the
key
for
this
problem.
It's
very
detailed
I
mentioned
discussed
in
this
issue.
149.32
the
the
the
root
cause
is
that
there
is
a
misuse
in
the
library
that
was
not
reuses
and
it
cannot
decode
a
key
that
was
encoded
when
it's
using
goal
114.
C
C
We
to
fix
this
issue
we
have
to
because
because
we
cannot
fork
notaries
code
and
bump
up
the
dependency
due
to
the
complexity.
C
So
we
decided
to
roll
back
to
use
goal
1.14
to
compile
notary
until
notary
fixes
this
issue
and
therefore,
if
you
are
using
harbor,
2.2,
0,
2.2.1
2.2.2
and
the
2.3.0,
and
you
created
a
key
when
you
are
signing
the
image
when
you
bump
up
to
2.2.3
or
2.3.1,
you
will
see
another
issue
same
as
before,
because
there
is
another
brick
change
we
introduced.
C
The
key
cannot
be
decoded
again,
but
we
believe
this
is
the
right
thing
to
do,
because
when
we
roll
back,
you
can
decode
the
key
that
was
created
in
the
earlier
versions.
Hope
you
are
still
following
me
and
to
work
around
that
we
mentioned
in
the
release.
Note
that
there
is
a
workaround
you,
you
will
have
to
remove
the
key
and
use
and
sign
the
image
again
such
that
the
key
can
be
regenerated.
C
So
this
is
the
usual
one
for
9
32..
In
short,
there's
a
break
change
we
introduced
into
that
2.3
and
2.3.
I
want
to
fix
this
issue,
so
I
I
hope
that
explains
the
problem.
D
C
D
C
C
Yep
and
the
other
issue
is
regarding
oidc.
This
was
about
introduced
when
we
are
refactoring
the
api
in
2.3.0.
C
The
symptom
of
the
problem
is
that
when
you
use
this
cri
secret
to
log,
in
the
I
mean
when
you
use
docker
cri
to
log
in
via
the
ci
secret
by
oidc,
it
may
happen
that,
when
the
token
is
refreshed
the
a
column,
a
sub
issue
is
flashed
to
empty.
C
Due
to
a
bug
and
give
me
a
second.
C
C
C
It's
still
under
reveal,
but
the
general
idea
has
been
verified
and
tested.
So
when
harvard
start
up,
it
will
try
to
get
the
record
whose
sub
issue
is
empty
and
try
to
decrypt
the
token
that
was
stored
in
database
and
restore
this.
C
C
I
I
I
think,
he's
scared
when
that's
weak,
if
yeah,
if
this
you
know
the
the
reveal
goes
fine,
and
we
have
this
all
the
fixes
merged.
C
Okay,
but
that's
only
for
the
oidc,
you
know:
authentication,
okay,
I'll
pass
it
to
you,
alex.
B
B
B
B
All
right,
yeah
thanks
daniel,
so
we
we
finished
2.3
and
we
released
2.3
and
so
we're
going
through
2.4
right
now.
B
So
if
you
head
over
to
the
project
board
right
from
the
main
project,
there's
a
project
board
here,
you
click
on
that
you
can
look
at
the
items
that
are
in
planning
for
2.4,
so
we're
trying
to
finalize
this
list,
and
so
I
wanted
to
talk
about
some
of
the
issues
that
we're
planning
to
to
tackle
in
2.4,
and
so
2.4
would
be
the
last
minor
release
of
the
year
for
us
all
right.
B
So
we're
looking
at
you
know,
sometime
before
kubicon
the
release,
we're
gonna
have
I'm
trying
to
have
an
rc
build
for
kubicon.
You
know,
as
we
usually
do,
and
then
the
actual
ga
will
come
out
a
few
weeks
after
that.
B
B
This
new
type
of
robot
account
that
you
create
under
system
admin
that
allows
you
to
configure
the
projects
that
this
robot
account
is
tied
to
right.
So,
instead
of
creating
it
under
the
project
admin,
the
system
admin
will
create
system
level
robot
accounts,
so
the
goal
is
always
to
deprecate
project
level.
Robot
accounts
at
some
point
and
just
only
have
a
system
level,
so
we're
gonna
we're
gonna.
B
We're
gonna
be
doing
this
in
2.4,
so
moving
forward
you're
only
going
to
have
the
system
system
level.
Robot
accounts
just
need
to
make
sure
you
know,
as
you
upgrade
those
existing
robot
accounts.
E
Yes,
can
you
hear
me
well.
E
Excellent
thanks
so
regarding
this
is
currently
a
project
administrator
is
able
to
create
project
level,
robot
accounts,
of
course,
and
that's
quite
handy
for
us,
but
our
system
level
robot
accounts.
E
It
is
possible
to
scope
them,
so
a
project
administrator
can
create
a
robot
account
that
can
act
on
scoped
to
just
the
project
that
the
project
admin
has
access
to,
but
not
create,
for
example,
well
privileged
accounts
for
other
things
than
the
particular
project
they
have
access
to.
Did
you
get
my
question.
C
C
Yeah,
could
you
please
open
an
issue
to
ask
the
question
because
the
guy
want
to
implement
this
feature
is
not
in
this
meeting.
However,
I
believe
the
project
level
robot
account
api
will
remain.
C
The
only
thing
that
will
not
work
in
2.4
is
the
v1
robot
account
whose
password
is
a
jwt
token.
If
you
have
that
kind
of
robot
account,
it
will
no
longer
work,
but
you
can
still
create
a
new
robot
account
on
project
level.
C
E
D
Yeah,
you
know
just
one
comment
is
here:
while
we
try
to
deprecate
the
latest
legacy.
B
Correct,
I
think,
yeah,
I
think
that's
that's
a
fair
request,
but
I
think
it
was
always.
You
know
the
goal
to
have
system
level.
Robot
accounts
moving
forward.
B
C
A
C
B
A
B
So
that's
the
first,
that's
the
first
issue
and
then
the
second
one
is
we're
going
to
try
to
add
some
kind
of
a
distributed
tracing
capabilities
to
harper.
So
you
can
trace
a
request.
This.
B
B
For
further
analysis,
I
don't
know,
if
is,
is
dun
chin
here
he
is
the
implementer
of
this
feature.
C
If
not,
but
we,
I
think
we
discuss
it,
but
one
thing
we
need
to
confirm
is
that
if
we,
you
know
implement
this
feature,
it
has
to
be
distributed.
C
Otherwise,
if
you
just
wanna
profile
the
performance
within
one
process,
there's
no
difference
if
you
just
enable
pprof
and
and
analyze
it
so
we
need
to,
you
know,
find
real
valuable
case
for
the
distributed
tracing,
but
yeah
I
mean
chen
has
did
some
good
poc,
so
we
have
figure
out
how
to
do
that.
We
just
need
to
find
valuable
use
cases,
because
you
need
to
do
a
lot
of
editing
to
code
to
inject
everywhere.
To
make
sure
you
know
it's
send
this
open
telemetry
data
around.
A
B
So
the
next
one
is
integrating
six
store
cosine
into
harbor,
so
harvard
today
uses
notary
an
open
source
project
called
notary
for
signing
images,
and
the
biggest
issue
with
notary
today
is
that
you
can't
replicate
a
sign
artifact
from
one
registry
to
another
right
so
from
harper
to
another,
another
docker
registry
like
harper
or
some.
You
know
other
cloud
registry,
perhaps
it
just
when
you
sync
an
image
to
a
different
instance.
The
signature
is
lost
right,
and
this
is
by
way
of
how
you
know.
B
Signatures
are
are
designed
in
notary
today,
right,
which
is
when
you
push
the
image
with
with
a
docker
content.
Trust
turned
on
it's
basically
just
an
ack
in
the
notably
db.
There's,
no
physical,
there's,
no
physical,
not
physical,
but
there's!
No,
you
know
the
signature
is
not
an
attribute
of
the
image
per
se
or
it's
not
something
that's
attached
to
the
to
the
image
right.
B
It's
just
it's
just
an
entry
in
the
data
database
and
you
know
notary,
has
been
working
on
the
sting
clinically
or
they've,
been
working
on
the
next
iteration
notary
called
negativity
v2,
which
is
which
is
for
this
multi-registry
type
of
setup,
and
so
we've
been,
I
mean,
we've
been
working
with,
we've
been
participating
in
those
discussions.
You
know
we've
been
following
along
for
quite
some
time
now,
but
you
know
it's
been
brought
to
our
attention
that
there's
another
project
called
cosign,
led
by
some
of
the
folks
at
google,
and
it's
just.
B
This
is
an
alternative
method
of
signing
images
right,
and
this
can
potentially
help
us
solve
that
request,
our
own.
That
request
of
you
know,
replicating
a
sign
artifact
from
one
registry
to
another
and
being
able
to
preserve
that
signature.
So
you
can
make
sure
that
the
image
you
replicated
is
indeed
the
same
image.
B
So
we're
not
going
to
be
that's
it
here.
This
is
the.
This
is
just
an
alternative
way
to
do
it
right,
we're
looking
we're
investigating
how
to
integrate
coastline
anti-harbor.
So,
as
a
user,
you
can
choose
the
designer
you
want,
but
yeah
I
mean
the
biggest
difference
here
to
my
recollection
is:
is
the
cosine
will
actually
produce
an
artifact?
B
B
So
you
know
the
same
sort
of
functionalities
that
we
have
today
right
around.
You
know
blocking
or
gating
image
poles
from
harvard
based
on
the
signature
profile
of
that
image.
A
B
There's
a
separate
artifact
from
the
image
itself.
You
know
a
lot
of
the
the
policy
management
capabilities
that
we
have
today.
How
would
how
would
those
apply
to
the
image
to
that
signature
right?
Those
are
some
of
the
questions
that
we
were
thinking
about.
C
Yeah,
I
I
want
to
mention
that
it's
possible.
We
only
have
a
concrete
design
when
two
diaper
is
released
and
we'll
see
because
there
are
so
many
parts
moving.
So
we
are
not
sure
if
we
can
finish
the
end
to
end
and
make
it
a
concrete
use
case
by
you
know
the
timeline
of
2.4,
but
we
will
try
and
just
stay
tuned.
I
think
jan
is
also
working
on
this.
He
will
update
with
some
progress
in
the
subsequent
meetings.
B
Yup
nope,
that
sounds
good
yeah,
so
we
don't
know
what
the
the
final
experience
is
going
to
be
like.
I
guess
it's
because
it's
so
new
the
project
is
very
it's
very
new
right
and
you
know:
we've
been
following
its
progress,
so
we're
just
kind
of
testing
it
out
and
I
guess
providing
alternative
if
I'm
aiming
and
we're
not
moving
away
from
note
3
right,
notably
v2,
is
still
ongoing
and
they're
they're
trying
to
solve
this
problem
as
well.
So
just
stay
tuned
and
find
out.
C
B
B
Yeah,
this
is
an
issue
with
the
existing
scans
within
harper
today,
which
is
you
know,
a
lot
of
operating
systems
that
can't
be
scanned
by
the
scanner
just
it
shows
up
as
where
is
the
specific
it
shows
up
as
having
no
vulnerabilities.
I
believe,
instead
of
you
know
accurately,
reflecting
that
it's
not
it's
not
covered,
or
it's
not.
B
Or
that
you
know
it's
not
really
recognizing
the
dos
or
the
distro,
and
so
you
know
we
have
to
work
with
the
scanner,
vendors
and
the
scanner
adapter
that
we
provide
to
make
sure
that
this
is.
This
is
fixed,
it's
just
something.
People
have
been
complaining
about,
and
absolutely
I
think
this
is.
This
needs
to
be.
B
B
So,
let's
see
here
replication
yeah,
so
this
is
basically
there's
no
way
to
in
the
current
replication
rules.
There's
there's
no
way
to
specify
a
list
of
exceptions
right.
Sometimes
you,
if
you
use
like
a
double
star,
for
example,
it
covers
everything,
there's
just
no
good
way
to.
C
B
Tell
harvard
not
to
replicate
a
specific
image
when
it
fits
when
the
tag
fits
the
the
filter
set
under
the
replication
policy
right.
This
is
just
another
additional
field
that
you
can
use
to
perhaps
specify
a
list
of
you
know
single
images
or
a
series
of
images
that
they
don't
want
to
be
replicated.
B
And
then
you
know,
there's
there's
a
larger
problem
that
there
are
various.
There
are
various
fields
here
within
harvard
today
right
for
different
for
different
policy
management,
so
replications
and
retention
policies
and
immutability
policies,
and
they
are
not
100
consistent
and
so
the
second
part
of
this
will
be.
B
You
know,
adding
making
sure
that
they
are
consistent
and
then
we're
going
to
be
finally
trying
to
to
to
add
support
for
regular
expressions,
proper,
regular
expressions.
B
So
you
can
use
everything
that
you
know
you're
accustomed
to,
so
this
is
going
to
be
a
fix
in
the
next
minor
and
then
the
regular
expression
support
will
probably
come
after
that
in
the
2.5.
I
think.
B
Yep
vulnerabilities
export
feature-
I
think
this
is
asking
for
you
know
way
to
export
vulnerabilities,
scanned
vulnerabilities
from
scandi
image,
scan
images
within
hardware
right,
a
single
click
on
the
ux
that
allows
you
to
export
everything,
so
this
is
going
through
a
ui
ux
design.
Right
now
we
don't
know
what
the
the
final
experience
will
look
like,
but
this
is
this
is
the
idea
right
and
I
think
you
know
I
think,
as
a
project
admin
right.
B
My
ideal
experience
would
be
if
I
have
a
lot
of
projects
within
harvard
today,
I
don't
want
to
have
to
go
into
every
single
project
and
export
as
a
separate
file
right.
I
want
to
be
able
to
one
click,
export
all
of
the
images
or
the
scan
results
of
the
images
that
I
have
access
to.
So
I
think
you
know
this
feature
whether
we,
whether
we
made
this.
B
B
So
yeah,
if
you
have
any
input
here,
if
you're,
if
you
you've,
been
using
harper
for
a
while
and
there's
a
specific
experience
you're
after
here
to
make
you
know
exporting
exporting
cv
data
much
easier.
Please
add
it
to
this
ticket
again.
You
can
find
everything
that
we've
discussed
today
by
looking
at
the
project
board
right.
So
again,
it's
here-
and
this
is
in
the
rightmost
column
right
now
called
2.4
product.
B
Anything
about
2.4
is
fine
yeah,
who.
E
Excellent
yeah,
I
do
have
an
question.
E
I've
read
about
your
your
goal
to
rip
to
deprecate
the
shark
museum
as
part
of
the
2.4
release
and
in
my
project
we
have
recently
started
using
harbor
and
we're
fairly
happy
with
it.
But
we
are
quite
worried
about
this
change
and
there
are
a
number
of
different
things.
Many
of
them
have
been
brought
up
in
the
discussions
on
github,
but
there's
one
point
that
I
think
is
particularly
important
that
I
want
to
to
get
some
response
on.
E
This
is,
I
think,
clearly,
breaking
change
to
the
api
people
that
have
implemented
workflow
today,
using
harbor,
pushing
images
or
pulling
images,
sorry
helm,
charts
from
harbor.
They
will
have
to
change
their
implementations
because
the
api
will
change.
So
this
is
indeed
a
breaking
change,
but
yet
you
include
it
in
a
minor
release,
so
I'm
wondering
either.
E
C
That's
a
good
argument,
I
I
have
to
say
and
yeah
I
I
think
we
hear
this
concern
and
comments
regarding
the
I
mean
from
the
community
regarding
removing
the
car
museum.
So
I
would
like
to
clarify
that
by
deprecation
it
doesn't
mean
that
we
are
going
to
delete
it.
C
C
Have
bandwidth
to
fix
it
and
the
maintainer
for
tar
museum
is
not
very
active,
so
yeah,
that's
why
we
want
what
we
we
really
did
this
discuss.
If
you
want
to
remove
it,
I
hope
we
can
remove
it,
but
yeah
I
mean
the
reality.
Is
people
still
love
to
use
it,
so
we
decided.
E
E
I
I
definitely
feel
you
it's.
It
would
be
such
a
nice
thing
to
have
oci
and
said,
and
I'm
not
going
to
go
into
the
other
issues
I
think
they
are
are
mentioned
in
a
fair
bit
in
the
discussions.
E
E
C
D
D
We
have
noticed
communities
still
allow
this
under
and
to
use
this,
so
we
have
already.
We
have
also
discussed
again
and
just
like
daniel
mentioned,
we
will
keep
this
function
still
here.
D
User
can
still
use
it,
but
we
will
with
limited
support
for
it,
and
we
also
will
come
if
in
case,
the
usual
fund
in
for
china
museum.
We
hope
the
community
can
also
work
on
those
issues
to
fix
that.
E
F
F
Go
ahead,
sorry,
and
for
you
from
your
perspective
I
mean:
how
long
does
it
does
it
take
for
you
to
move
to
the
oci
chat,
support
from
your
perspective,
if
you
want
to
move,
are
you
or
you
don't
want
to
move
it
completely.
E
Well,
we
we
definitely
are
excited
to
move
to
oci,
but
our
customer
is
very
keen
on
using
stable,
released
features
and
oci
support
in
helm
is
not
stable,
yet
it's
still
experimental.
So
we
cannot
implement
it
fully
for
our
customer
before
helm
have
implemented
it
once
they
have
implemented
it.
I'm
sure
we
will
be
able
to
solve
it
in
a
matter
of
weeks,
not.
A
E
F
E
F
We
have
discussed
previously
that,
because
we
have
deprecated
the
chatter
museum
right,
so
we
would
like
to
put
more
effort
to
fix
the
helm
itself
to
spot
oci.
F
E
Yeah,
that's
great
we're
happy
to
contribute
when
there's
things
that
we
can
can
help
with
it's
great
to
hear
that
you're
open
for
contributions
and
I'm
looking
forward
to
to.
B
E
C
B
You
can
look
up,
there's
an
issue
in
the
repo
that
basically
it's
an
issue
with
chart
museum
and
we,
like
daniel,
said
we
were
not
able
to
get
properly
supported
on
an
upstream
right
and
we've
had
this
experience
with
docker
distribution
previously,
as
well
as
with
notary.
B
We,
if
we
you
know,
if
we're
depending
on
something
that
doesn't
get
the
right
amount
of
attention,
we
have
to
lean
ourselves
off
it.
Basically,
we
have
to
become
don't,
seek
another
solution
or
become
more
independent,
that
right
and
and
so
that's
kind
of
the
the
problem.
The
tribe
museum
is
not
like
a
a
trivial
one
right,
it's
it's.
B
As
you
scale
the
number
of
charts,
it
has
performance
issues
and
we're
not
able
to
get
a
conversation
started
even
right,
we're
not
getting
any
kind
of
feedback
on
the
issue,
and
so
I
know
that
this
is
a
big
change
right,
especially
for
people
like
you
said
right,
especially
for
anyone
that
has
built
products
around
it
and
they're
leveraging
it
pretty
heavily.
But
yeah
I
mean
we
basically
weighed
the
pros
and
cons
of
each
and
so
that's
kind
of
the
direction
forward.
B
Said
it'll
still
be
there
for
2.4.
Perhaps
I
wasn't
too
sure
on
that,
but
I
think
thank
you
daniel
for
clarifying
that,
but
you
can
still
use
it
in
2.4,
but
it's
not
going
to
be
it's
not
going
to
be
supported
very
well
by
the
harper
team.
So
we
have
to
make
sure
we
get
this
message
across
in
the
release.
Notes
as
well.
A
A
Maybe
not
I,
I
have
a
general
discussion
to
add
about
the
community.
There
was
a
idea
to
stream
the
community
hours,
the
community
meetings
and
the
office
hours
via
youtube,
because
people
are
not
feeling
okay
joining
in
actually
actively
participating
in
the
meetings.
So
what
do
you
think
folks
is
that
makes
sense,
so
we
can
stream
and
at
the
same
time
join
over
zoom
if
you
want
to
actively
participate
in
the
meeting.
B
I'm
okay
with
it
yeah.
B
C
C
B
A
But
in
our
case,
that's
like
more
like
a
real
conversation
that
we
have
right
now
and
the
kgi
is
more
like
a
single
person
show
right.
A
Yeah,
but
if
they
want
to
ask
questions,
we
have
to
encourage
them
to
join
the
zoom
and
to
be
real
participants
and
youtube
to
be
only
for,
like
a
read,
only
viewers,
if
it's
more
accessible
for
someone,
so
it
was.
I
idea
that
I
think
valero
are
doing
this
for
for
a
while,
and
you
can
give
it
a
try
if
the
community
feels
okay
with
this
one
and
see
how
it
goes
just
is.
C
D
Yeah,
thank
you.
Hey
we
yeah.
We
still
have
an
agenda
today.
Can
you
open
that
agenda?
We
have.
We
have
daniel
monego
from
aurora
that
originally
worked
on
the
armor
arm
release
for
hubble,
or
we
invite
him
to
provide
some
update
for
the
hubble
arm.
C
D
Yeah
no
link,
I
think
daniel
monega
will
provide
some
update
on
the
update.
H
Okay,
okay,
so
nice
to
be
here
again,
it's
been
a
few
few
months
since
our
last
update,
so
I'm
coming
today
to
bring
some
updates
on
the
multi-arc,
the
kind
of
proposal
and
on
on
the
arm
release
side.
First,
first
of
all,
like
from
the
last
time
we
talked
for-
or
at
least
we
updated
things
here,
we
had
a
few
questions
on
the
proposal.
Actually,
the
proposal
is
still
open.
H
H
Oh
so
bad.
I
still
have
some
problem
here
with
references
but
I'll
try
to
share
it.
H
Sure
I'll
put
the
link
on
the.
H
So
this
is
the
yeah,
isn't
it.
This
is
the
proposal.
H
Okay,
so
so
this
is
the
proposal
we
actually
changed
a
little
bit
from
the
beginning
towards
more
like
a
maturity
level
or
how
new
architectures
are
integrated
into
harbor.
There's
still
lots
of
comments
in
there.
It's
a
little
bit
not
have
so
much
activity
here,
but
if
anyone
has
any
any
questions
or
has
any
insights
or
ideas,
you're
welcome
to
to
add
new
comments.
I
think
the
most
gross
part
part
of
the
proposal
is
ready,
so
we
like
to
to
push
it
to
to
integrate
as
soon
as
possible.
H
We
have
discussed
a
little
bit
because
from
this
first
time
we
pushed
this
change.
We
we
changed
a
little
bit
the
direction
towards
this,
like
I
said,
like
a
more
maturity
level
kind
of
process
and
how
to
integrate
new
architecture,
so
it
would
be
really
nice
to
have
more
more
folks
like
looking
into
it.
H
The
second
part
we
want
to
to
update
today
is
that
I'm
sending
the
second
link
we
have.
We
already
started
to
work
on
how
to
build
arm
arm
64
releases,
it's
still
a
working
process.
There
are
a
few
things
that
we
are
still
adjusting
in
the
rep
repository.
We
have
a
pull
request
here
that
we're
still
working
on,
but
we
can
already
build
arm
arm
images
it
it
works.
We
we
tested
it
on
our
servers
and
it
works.
Okay.
H
We
still
have
one
issue
that
is
not
solved.
That's
regarding
redis
redis
has
a
kind
of
an
issue
using
photon
photon
as
a
base
image
that
when
running
an
arc
in
arm
it
doesn't
it
doesn't
work
very
well.
So
we
with
in
our
local
tests,
we
changed
to
use
the
redness
library
image
and
it
works
okay,
and
we
already
we
already.
I
see
that
there's
an
effort
to
to
talk
with
the
photon
members
to
to
fix
this
issue
and
we'll
keep
posting
updates
as
soon
as
we
have
more.
H
This
is
a
blocking
issue
for
us
right
now
and
since
this
one
once
this
blocking
issue
is
fixed
and
this
pull
request
is
is
merged.
All
the
changes
are
are
okay.
Then
we
will
start
to
work
on
automate
like
changes,
automate
testing
for
for
arm
releases
and
once
that
that
infrastructure
and
the
the
process
is
all
built
up,
and
we
will
start
to
to
to
to
brings
up
to
speed
with
the
with
the
upstream
of
md
md64.
H
H
B
Hey,
thank
you
daniel.
I
have
a
question
on
it.
Can
you
perhaps
share
you
know
some
data
on
the
usage
of
harbor
for
arm?
What
is
the
expected
usage
and
just
what
kind
of
you
know
deployments?
Are
you
looking
at?
Why
is
this
important.
H
Yeah,
so
you
know,
arm
is
becoming
like
more
more
and
more
mainstream
outside,
like
you
can
see
now
you
can
deploy
servers
using
aws
and
other
and
other
cloud
providers,
and
it's
getting
up
to
speed
and
maybe
even
for
some
kind
of
using
raspberry,
pi
or
some
local
development
is
also
very
useful.
H
We
currently
in
alauda
we
we
can.
We
have
a
kind
of
a
big
demand
on
on
harbor
as
distributed
as
an
arm
artifact.
We
currently
we
build
this
internally.
We
kind
of
did
change
a
little
bit
of
the
source.
Did
some
source
code
change
to
build
it
internally
and
that's
why
we're
actually
contributing
to
make
it
into
the
as
an
official
package?
H
Currently
I
I
don't,
I
cannot
say
for
sure
the
numbers
exactly,
but
the
demand
is
quite
high
for
us
and
we
expect
to
once
this.
The
all
those
issues
are
fixed
and
all
these
blocking
issues
are
solved.
We
expect
to
to
release
a
2.3
release
and
from
them
from
them.
H
B
Okay,
thank
you,
and
do
you
want
to
just
you
know,
introduce
maybe
say
a
few
words
about
a
lotta
and
how
you're
using
arm?
I'm
not
sure
everyone
here
knows
a.
H
Lotta,
oh
sure,
so,
lauda
is
as
a
company
in
china.
Here
we
we
work
with
container
technologies
to
to
help
our
customers
to
embrace
cloud
native
technologies.
We
have
also
a
few
other
projects
in
the
cncf
community,
like
qovn,
which
is
a
cni
as
a
network
plug-in
for
kubernetes,
and
we
still
have
some
other
products
in
line,
and
we
expect
to
have
a
few
more
projects
open
sourced.
Soon
we
work
generally
with
kubernetes
and
fibonaccis
based
solutions
in
kind
of
platform
as
a
service
product.
D
Yeah,
thank
you,
daniel
just
one
question
is
there
any
I
mean
rough
date
for
already
for
the
arnold
is
2.3.
H
Yeah
we
actually
we're
trying
to
talk
about
this
today.
Internally,
we
have
a
few
kind
of
urgent
tasks
that
we
are
having
to
deal
with
internally
first,
but
we
we
expect
to
to
do
to
have
it
done
soon,
and
then
we
can
focus
again
back
on
on
the
arm
release.
We
expect
to
I
I'll
I'll
give
a
shot
on
a
date,
but
we
expected
to
be
late,
maybe
in
beginning
of
the
next
month,
maybe
august,
beginning
of
august.
A
You
and
one
one
thing
for
me:
daniel,
do
you
think.
G
A
A
You,
okay,
I
think
that
was
the
last
item
from
the
agenda
to
today.
Anyone
else
last
minute
call.
A
H
Question
actually,
regarding,
like
the
robot
accounts,
I
saw
that
we're
talking
about
sort
of
product
accounts.
Today
we
kind
of
have
a
small
request
regarding
robot
accounts.
I
know
that
currently
it's
only
useful
to
pull
and
push
artifacts,
but
it
cannot
access
the
api.
Is
that
correct.
B
When
you
create
well,
it
has
limited
access
to.
You
know
a
subset
of
the
api
that
we
believe
are
you
know
enough
for
the
robot
accounts
for
for,
but
it's
in
you
know
it's
what
it's
intended
to
do,
and
so,
when
you
create
the
robot
account
and
their
system
view,
you
can
select
the
api
actions
and
you
can.
You
can
modify
those
later
as
well,
but
if
you're,
if,
if
something
that
you
need,
if
it's
an
api
that
you
need
and
if
that's
not
covered
today,
definitely
feel
free
to
log
an
issue.
A
Yeah,
thank
you.
Everyone
for
attending
and
see
you
next
time
and
we
have
the
office
hours
today.
Is
that
true,
no
no
officers.
A
G
So,
yes,
we,
we
might
bring
them
back
later.
This
fall
with
the
with
the
intent
of
doing
more
deep
dives
into
issue
discussions
and
technical
documentation.
G
G
Alternating
weeks,
so
we
have
the
community,
we
meetings
bi-weekly,
so
we
would
have
the
office
hours
at
the
same
time,
but
during
the
off
weeks.
So
we
would
have
this
time
every
week.
It
would
either
be
a
community
meeting
or
an
office
hours.