►
From YouTube: CNCF Harbor's Community Zoom Meeting - May 3, 2023
Description
CNCF Harbor's Community Zoom Meeting
A
Recording
in
progress,
I
suppose
anyway
welcome
everyone,
and
today
is
May
the
3rd
and
that's
official
community
meeting
for
Harbor.
My
name
is
Olin
and
I'm
the
community
lead
for
Harbor.
Thank
you.
Everyone
for
joining.
Please
follow
the
cone
of
conduct,
as
this
is
official
cncf
meeting
so
yeah
with
that
I'm
gonna.
Send
you
the
chat,
our
community
meetings
notes
into
GitHub.
So
if
you
like,
add
yourself
there
at
your
topic,
so
we
can
go
through
them
and
discuss
them
for
today
and
I'm
gonna.
A
A
B
A
B
B
So
it's
planned
for
next
week
we
will
see
how
it's
going
and
yeah
nice
yeah
and
regarding
the
terraform
provider,
so
I
have
a
backup
from
my
employer.
So
we
can.
We
can
be
part
of
it
so
to
say
with
our
work
time.
So
it's
no
problem
check
that
out,
but
so
check
the
repository
out.
But
it's
yeah.
It's
here.
B
A
A
A
D
A
D
A
A
A
A
F
Of
course,
and
one
of
my
colleagues,
the
elephant
was
also
in
the
in
the
Harmon
Harbor
Meetup,
which
turned
out
to
be
quite
a
small
room
if
I
remember
correctly
so
I
hope.
Next
time
more
people
can
join,
but
maybe
not,
but
yeah
quite
an
interesting
experience
and
yeah.
We
also
use
harbor
on
our
premises.
We
also
now
do
the
integration
with
geosob
so
make
it
high
available
with
replication
of
projects,
because
yeah,
redundancy
and
I
was
wondering
what
kind
of
efforts
are
ongoing.
F
Some
kind
of
thing
you
can
use
to
talk
to
a
harbor
because
I
remember
in
the
past
we
had
issues
with
functionality
which
was
not
available
in
the
GUI
at
the
time,
and
I
made
a
small
client
to
to
do
stuff
like
removing
the
old
repost
the
how
you
call
that
functionality
to
get
rid
of
all
the
versions,
but
I
just
was
wondering
what
kind
of
efforts
are
going
in
that
direction.
I
don't
want
to
duplicate
work.
F
A
You're
dropping
out
a
little
bit
so
I'm,
not
sure
if
I
got
it
everything
but
I
suppose
you
were
asking
about
the
CLI,
yes,
so
to
prompt
you
in
this
one.
We
we
applied
for
the
links,
Foundation
mentorship
program
and
we
have
one
of
the
project.
Is
that
CLI
thing
so
a
chat?
Who
is
right
now
not
on
the
call?
But
he's
he's
working
on
that
on
that
effort
to
streamline
and
make.
A
A
A
A
D
Already
popped
up
here,
but
I
I,
don't
come
yeah,
so
I'm
here
concerned
I'm
the
only
one
managing
the
registry
here.
So
there's
that
and
I'm
trying
to
just
here
to
see
what
you
talk
about
and
that's
it
I'm
currently
I'm
most
active
in
the
hardware
itself
in
the
repository
where
I
try
to
fix
my
problems
and
understand,
what's
going
on
with
people
and
sometimes
I
commit
something
back
mostly
for
the
health
chart,
though
so
yeah.
A
Nice
thank
you
and
welcome
yeah
I'm
gonna
ping,
you,
if
that's
okay,
and
that
that's
also
valid
for
Ono
I'm
gonna,
bring
you
offline.
If
you
can
share
a
little
bit
of
your
story,
I'm
trying
to
collect,
like
the
overview
of
who
is
using
Harbor
and
how,
if
it's
possible,
to
share,
of
course,
so
so
that
can
be
like
I'm
not
trying
to
build
like
a
community
knowledge
kind
of
Hub.
But
it's
more
of
a
effort
to
streamline
a
little
bit
different
use.
A
F
A
F
Right
or
you
could,
of
course,
add
local
hip
to
the
main
first.
That
was
like
the
major
reason
to
to
massively
start
using
Harbor
because
of
the
change
in
policy
for
Docker
hub,
and
there
were
some
some
I
think
there
were
some
hacks
in
the
past
as
well
over
I
forget.
But
there
was
also
this
this
integration,
which
was
no
longer
freely
available
and
things
were
getting
hard
and
yeah.
A
F
A
A
G
C
A
A
G
Yes,
okay,
so
this
is
the
proposal
for
adding
more
fine
grained
versatile
replication
filters.
So
these
are
like
the
goals
for
adding
it.
So
initially,
the
meditation
I
had
in
mind
was
to
support
the
semantic
version
numbers,
because
the
initial
issue
has
the
requirement
for
supporting
your
the
version
numbers.
G
So
that's
what
I
initially
thought,
but
then
after
I
started
the
mentors
we
figured
that
should
support
like
ad
or
give
the
option
to
the
users
to
specify
the
regular
expression
by
themselves
so
that
they
can
have
a
complete
control
of
the
replication
rules.
So
these
were
the
initial
two
approaches
so
of
this.
So
in
this
whatever
the
second
approach
that
I
had
in
mind
was
that
to
accept
a
regular
expression
from
the
user
on
the
portal
and
then
and
perform
the
filtering
based
on
that.
G
So
this
is
just
a
regular
expression
for
the
semantic
version
number
thing,
and
apart
from
that
last
week,
I
had
a
chat
with
Vadim
and
we
figured
out
four
to
five
ways
of
how
to
get
the
expression,
the
regular
expression
from
the
user.
So
one
of
that
is
adding
in
new
input
field
for
a
regular
expression.
So
right
now,
if
you
see
the
replication
rules,
why
these
are
the
four
options
available
here
and
the
we?
The
approach
proposes
to
add
a
separate
field,
and
that
would
just
take
input
to
the
regular
expression.
G
So
this
approach
they
couldn't
need
any
migration,
because
since
the
regular
expression
rule
would
be
added
as
a
red
field,
the
rules
that
are
existing
would
still
work
and
no
migration
would
be
needed
and
users
can
switch
to
the
regulation
based
filtering
as
their
convenience.
But
it
has
a
development
and
logical
complexity
because
the
users
might
try
to
buy
in
the
regular
expression
based
Rule
and
the
existing
double
star
pattern.
Rule
and
it
might
cause
some
conflicts
in
there.
G
So
it
has
its
quantities
and
the
changes
are
also
needed
on
the
UI
portal,
the
API
handlers
and
the
filtering
logic.
So
it
is
time
consuming
and
too
many
moving
parts
to
handle
and
or
need
to
change.
So
this
is
not
quite
preferred,
then,
except
regular
expression
in
existing
input
fields
and
distinguish
between
the
glob
pattern.
G
The
regex
pattern,
so
the
existing
input
fields
that
are
there,
the
regular
expression,
can
be
accepted
in
here,
along
with
the
existing
rules
and
to
distinguish
whether
a
user
is
trying
to
match
the
glob
pattern
or
the
regular
pattern.
It
could
be
specified
in
the
forward.
Slash
splashes,
just
like
we
write
normal
regular
expression
to
distinguish
and
the
distinguishing
and
we're
done
on
the
back
end.
So
no
change
would
be
needed
on
the
UI.
G
Just
on
the
back
end,
we
would
have
to
distinguish
how
to
identify
a
regular
expression
and
the
pros
or
use
of
existing
UI,
and
the
existing
filters
do
continue.
Working
but
again
the
same
problem
that
users
might
try
to
combine.
The
glob
and
regex
pattern
which
will
again
get
complicated
and
distinguishing
between
a
blob,
Rule
and
a
expression
rule
might
fail
on
the
back
end
due
to
unidentified
edge
cases.
G
So
it
needs
some
effort
on
the
testing
side
too,
and
database
schema
change
needed
to
Source
the
glob
tool
and
regular
expression
rule
separately,
because
one
field
might
be
the
regular
expression
field
and
another
might
be
the
globe
field.
So
the
even
this
is
not
what
I'm
personally
am
not
preferring
this
approach
and
then
the
third
approach
is
to
provide
an
option
for
choosing
between
the
glob
and
the
regular
expression
pattern.
G
So
a
radio
button
based
Choice
can
be
occurred
between
the
glob
pattern
and
the
regex
pattern,
and
once
the
regular
expression
pattern
is
selected,
then
the
user
can.
This
is
there
would
be
Fields
here
whether
the
user
wants
to
filter
the
name,
tag
or
label,
and
the
user
can
only
input
a
regular
expression.
So
it
is
only
regular
expression
and
only
glob
no
combination
of
both.
So
this
would
be
easy
to
understand
and
use
and
seamless
and
convenient
migration.
G
But
again
too
many
moving
parts
to
change
the
UI,
the
API
handles
and
the
database
schema
with
the
filtering
logic,
and
it
would
be
higher
maintaining
and
development
complexity.
Because
then
the
block
pattern
needs
to
be
made
and
the
regular
expression
pattern
needs
to
be
maintained.
So
that's
there
and
then
the
fourth
approach
is
to
completely
deprecate
the
Glock
pattern
and
migrantly
completely
to
the
back
end.
So
this
would
be
instant
migration
to
regex
rule,
providing
more
A.
G
G
This
one
might
not
quite
well
preferred
and
has
a
higher
rate
of
breaking
existing
installations
and
causing
migrations,
and
then
the
fifth
approach
is
accept
both
a
regular
expression
and
glob
pattern
and
UI
and
Implement
only
regex
based
filtering
in
the
back
end
and
then
convert
the
block
based
rules
to
regex
rules
in
the
back
end.
So
I'll
explain
this
one
in
a
better
detail.
So
what
we
will
do
is
on
the
UI
portal.
We
would
accept
both
the
Glock
pattern,
based
rules
and
regular
expression
rules.
G
On
the
back
end,
we
would
remove
the
Glock
pattern
filtering
the
filtering
logic
that
currently
exists
and
completely
implement
the
regular
expression
based
filtering
and
convert
these
glob
rules
into
regular
expression
rules,
so
that
only
filtering
logic
that
is
there
is
of
the
regular
expression
and
the
globbing
is
removed
and
users
don't
notice.
The
change.
Users
can
add
the
regular
expression
and
the
glob
patterns
and
whatever
glove
pattern
they
had
similar
to
the
existing
one.
G
It
would
get
converted
into
a
regular
expression
pattern,
and
so
existing
glob
rules
in
the
currently
which
is
stored
in
the
data
can
be
converted
to
regular
expression
rules
directly
on
the
runtime.
So
there
is
a
need,
and
apart
from
that,
a
migration
script
can
be
provided.
We
can
provide
a
migration
script
while
updating
to
the
newer
version
that
would
automatically
convert
the
existing
rules
into
regex
rules
and
store
in
the.
D
G
Database
and
existing
rules
continue
working
and
the
filtering
logic
gets
implemented
without
any
change
to
the
UI,
but
it
has
again
a
chance
of
breaking
existing
like
a
certain
upgrade.
It
might
break
some
installations
if
there
is
too
much
of
a
version
gap
between
the
user's
current
installation
and
the
newest
release
that
we
publish
so
it
might
break
and
also
need
the
migration
script,
which
is
another
moving
part
to
handle,
and
the
conversion
of
the
glob
pattern
to
regular
expression.
G
Pattern
must
have
a
hundred
percent
coverage
so
right
now,
I
don't
to
have
a
big
absolute
idea
of
how
what
the
various
ways
users
are
using
the
a
replication
filters.
So
there
might
be
some
unidentified
H
cases
that
might
cause
a
replication
rule
to
you
know
fail.
So
this
is
the
full
proposal
and
you
books
can
just
read
it
once
and
drop
in
your
thoughts
in
the
discussions,
and
we
would
be
happy
to
get
along
from
there.
So
the
fifth
one
is
what
currently
I
am
favoring,
because.
A
E
Okay,
when
the
the
we
we
will
have
a
lot
of
work
in
the
background,
but
we
still
I
still
personally
like
we
can
have
a
filter
mode
I'd
like
users
to
choose
which
one
he
wants
to
use
and
we
just
introduced
another
one.
And
if,
if
user
want
used
existing
the
double
star,
he
still
can
use
it,
and
if
we
have
another
of
another
option
and
we
do
have
a
application
policy.
So
if
we
wanted
she
migrate,
all
the
existing
policy
to
register
filter
amount.
E
We
can
follow
the
application
policy
that
maybe
we
can
fabricate
the
depth
star
mode
in
somehow
any
in
some
future
releases,
but
but
I
personally,
like
the
third
one,
the
other
comments
in
the
football.
So
thank
you
for
other
work
and
we
should
let
more
maintenance
to
review
this
of
proposals
to
add
their
opinion
right.
G
A
A
C
Okay,
hi
so
I'm
joining
from
VMware
and
I'm,
a
new
product
manager
for
Harvard,
so
I'm
trying
to
bring
understanding
on
Hardware
use
cases,
different
use
cases
and
what
I
get
to
hear
off
late
very
recently
about
the
application
of
chart.
Museum
looks
like
there
have
been
many
customers
who
have
been
using
this
feature,
especially
in
Harbor
OVA
case
and
their
team,
to
understand
if
this
can
be
still
used
and
what
is
a
reason
behind
application.
Despite
the
usage
and
the
utility
of
the
feature,
they
are
just
curious
to
no
problem
Community
standpoint.
H
A
E
Yeah,
this
is
the
background
of
why
we
decided
to
deprecate
the
Trump
Museum.
So
so
the
original
idea
was
that
the
the
chart
is
in
the
Upstream
charm
Museum
project
as
not
under
maintain
the
they
were
seeking
a
mintender
from
community
right
now
so
and
about
a
Helm
and
Harvard
team
believed
that
the
OCA
a
compatible
chart
are
the
future.
So
we
decided
to
move
from
China
Museum,
206
compatible
charts
and
we
announced
the
application
and
2021
yeah
about
one
and
a
half
years
ago.
So
via
KU
hello.
C
E
D
E
We
explain
why
we
wanted
to
remove
the
charm
Museum
and
allow
discussion
around
this
topic,
and
we
input
some
background
in
this
discussion.
Why
we
do
this
and
when
we
do
this
and
how
to
migrate
a
chart
and
files
into
the
OCR
compatible,
charts
and
yeah.
This
is
all
the
you
can
go
through
the
the
the
message
I
input
in
the
gay
heart,
and
it
has
a
lot
of
information
that
we
are
not
announced
to
the
community
and.
A
E
E
I
think
it
is
the
official
release
for
the
tensor
customers,
so
it's
out
of
scope
of
a
Harbor
community
meeting.
So
if
we
want
to
discuss
the
horrible
way,
I
think
you
can
contact
us
through
the
internal,
a
slack
Channel
and
it
is
now
maintained
by
the
Harbor
OSS
Harbor
community
and
it's
a
internal
product
within.
So
if
you
have
any
questions
regarding
to
the
horrible.
C
E
E
A
E
E
So
we
wanted
to
do
an
enhancement
on
this
and
right
now
we
are
working
on
that
and
we
want
to
optimize
the
the
whole
process
and
reduce
to
from
10
step,
maybe
to
three
or
four
stuffs
and
have
user
to
to
work
all
the
existing
travels
into
to
OCR,
charts
and
one
time
rather
than
one
time
for
a
specific
project.
So
well
probably
we
can
finish
it
and
finally
end
up
Network.
E
B
No
I
don't
think
so.
This
is
what
you
can
do
with
your
own
hand,
charts
and
I
think
this
is
quite
a
feasible
way
for
for
the
teams
and
transparent.
But
what
we
are
looking
for
is
and
there's
this
artifact
hub.io,
where
another
vendors
are
providing
their
hand,
charts
and
a
lot
of
our
teams
are
relying
on
that.
B
A
We
just
not
sure
when
exactly
you
joined,
but
Mino
asked
about
the
the
ham
chart,
the
chart,
Museum
deprecation
and
migration
and
stuff,
but
it's
just
more
related
on
VMware
internal
stuff,
with
the
OVA
shipped
with
the
tanzo
and
then
onion
shared
that
the
thing
which
I
for
some
reason,
I
missed
out
from
the
release
notes
it's
super
well
hidden
in
the
release,
notes
I
think
we
have
to
edit
a
little
bit.
The
list
notes
so
can
be
more
emphasized
on
this
steps,
because
people
are
like
super
frustrated.
A
D
A
Yeah
I'll
fix
the
release,
notes
and
actually
I'm
gonna
send
that
link
to
the
mailing
list.
If
someone
needs
to
to
follow
that,
and
they
miss
that,
it's
surprisingly,
how
many
folks
are
not
reading
their
lease
notes,
yeah,
including
me,
obviously,
I
I'm,
definitely
sleep
this
one
I,
don't
know
how
that
happened
and
I
wrote
the
blog
post
about
the
release
so
yeah
yeah
anyway.
It's
let's
not
getting
more
embarrassing
to
me.
Anyone
else
wants
to
share
anything
for
today.
F
Something
in
the
chat
chat
yeah.
Basically,
we
want
to
enable
how
you
say
that
this
allow
insecure
image
you
can
configure
Harbor
for
that,
but
it
doesn't
work
if
we
use
it
as
a
proxy
cash
to
Docker
Hub,
because
scanning
is
not
synchronous,
I
think
it's.
The
scanner
will
only
be
effective.
The
second
time
you
pull
the
image,
not
the
first
time.
H
Can
I
can
I
rephrase
this
question
again
because
I
think
it's
it's
already
on
the
issue,
so
you
mean
that
when
you
have,
and
so
when
you
enable
Harbor
as
a
proxy,
for
example,
for
Docker
Hub
right
yeah,
and
then
you
pull
the
image
from
the
proxy.
You
know
from
Docker
hub
and
this
image
is
pulled
and
you
can
pull
it,
but
once
you
put
when
you
pull
it
once
the
the
scanner
is
triggered
in
the
background
and
might
flag
the
image
as
vulnerable.
H
F
H
C
H
B
B
And
we
are
mirroring
certain
versions
and
x
and
chars
and
scan
them
and
have
them
secured,
so
the
teams
have
access
to
up
to
images
from
from
from
Reddit
repository.
But
we
have
this
kind
of
yeah
deployment
security
enabled
for
this,
for
this
images,
so
they're
not
going
to
production
without
scanning,
but
it
only
works.
If
you
don't
need
access
to
the
full
catalog,
because
you're
not
coming
to
manage
that.
C
A
Screen
we
we
fixed
that
okay,
fixed
is
not
the
right
term,
but
we
bypassed
that
issue
by
like
preheating.
So
we
knew
what
What's
going
to
be
put
from
the
from
the
internet.
Let's
say,
and
we
people
that
we
scan
it
and
eventually
we
fix
some
of
these
issues
or
not,
and
we
took
decisions
with
the
OPA.
How
how
we're
going
to
work
for
what,
with
the
images
from
from
GitHub
in
particular,
yeah
but
I'm,
not
sure,
if
that's
always
the
yeah,
workable
and
manageable
yeah.
B
We
are
doing
this
via
replication,
so
we
are
replicating
certain
images
into
one
of
our
projects,
scan
them
and
have
the
deployment
security
active.
So
if
the,
if
a
new
image
is
replicated-
and
it
will
be
scanned-
and
this
is
synchronous-
it
will
not
be
able
to
pull
at
the
first
time
if
it
has
some
certain
severity
in
their
vulnerabilities.
Okay,.
B
F
A
Yeah,
it
sounds
like
a
joke.
It's
almost
happening
once
yeah,
it's
a
bad
joke
anyway,
but
yeah,
but
if
you
so
we
can,
we
can.
We
can
take
a
look,
and
maybe
we
can
try
to
find
resources
to
put
some
priority
on
this
one
for
the
next
release.
If
it's
a
obviously
it's
a
very
old
issue
and
you're
not
fixed
so
yeah.
H
Yeah
and
I
think
it
will
be
difficult
to
find
a
solution
for
that
right
because
it's
you
know
the
conceptually
you're
proxying,
something
you
don't
know
you're
processing
right.
So
unless
you
only
know
it
when
you,
when
it's
done
right
and
then
it's
not
proxy
right,
so
otherwise
it
should
be
some
sort
of
a
proxy
relay
or
something
so
it
first.
When
you
request
something,
you
expect
that
the
data
is
you
know
coming,
but
yeah
I
think
it
will
not
be
solved.
I
mean
we
can.
A
A
D
B
D
A
C
A
B
That's
the
problem.
I
just
said
that
that
I
got
the
invitation
from
you.
So
thanks
for
that
very
kind
and
I've
planned
with
my
team
for
the
next
week,
a
little
Dojo
or
hack
session
to
make
this
work
in
our
environment
and
hopefully
provide
some
feedback
and
hopefully
have
then
a
good
decision
point
to
go
forward
to
2.8,
because
we
rely
on
handshot
museum
right
now.
But
we
want
to
get
rid
of
it.
G
H
Okay,
yeah
so
I
I'm.
Currently
you
know
writing
the
documentation
for
the
project,
how
it's
working,
how
to
spin
up
yourself
and
also
how
how
how
it's
working
and
then
the
inner
works
and
I
hope
I
can
present
it
in
two
weeks
in
more
detail
and
explaining
how
it
works.
But
basically
it's
a
proxy
right.
So
when
you
only
can
you
open
up
the
example,
you
have
the
test.
First
yeah,
if
you,
if
you
take
a
look
at
the
test
line,
you
know
like
the
yeah.
A
H
D
H
It's
fetching,
then,
from
chat
means
your
charts
it
now.
We
can't
pick
now
the
airflow
and
it
will
fetch
the
latest
one,
and
then
you
can
specify
the
concrete
where
the
the
exact
version-
and
this
is
actually
the
use
case,
for
it
solves
the
problem
that,
if
you're
relying
on
public
charts,
you
know
in
your
organization,
you
can
use
this
one
to
replicate,
create
replication
rules
that
will
replicate
charts
from
you
know,
public
endpoints
to
your
internal
organization,.
B
H
H
B
B
H
H
H
D
H
A
H
B
H
This
is
demo
instance,
and
we
have
here
charts
big
dummy
charts
right.
So
there's
like
350
megabyte
of
charge,
so
it's
basically
a
whole
different
number
charts.
We
copy
them
for
testing
and-
and
we
have
the
jet
stack,
so
the
search
manager
charts
as
well
here.
So
it's
also
like,
like
1000
charts
and
you
just
create
a
perfect
replication
rule
and
then
sorry
and
then
you
create
a
replication
Rule
and
then
you
replicate
from
sorry.
H
D
H
H
Yeah,
so
for
the
next
session,
I
will
create
a
demo
and
also
explain
how
it
is
working,
and
this
can
be
also
integrated
into
into
Hardware.
If
it's,
if
it
makes
sense
because
then
it
would
make
you
know,
consume
chart
chart
music
chart
charts
easily.
It
will
not
serve
charts
right
like
chart.
Museum
did
so
it's
not
a
solution
for
that,
but
it
will
make
it
easier
to
consume
charts
that
are
published
in
either
oci
format
or
a
check.
Museum
format.
E
H
E
H
E
Yeah,
sorry,
no
yeah
yeah,
but
currently
the
point
for
user
is
they
want
to.
They
want
us
to
provide
it
to
to
have
them
to
migrate,
existing
Chapel
right
of
Harvard
right
now,
so
maybe
written
Consolidated.
Our
existence
yeah
yeah
into
why-
and
there
are
two
for
use
it
to
to
migrate
either
existing
the
charts
or
or
replicate
the
the
the
phenomy
charge
to
convert
them.
Yeah.
H
This
this
can
be
easily
integrated
and
I
had
actually
the
idea
for
the
plan
to
create
to
have
it
not
just
running
as
a
service
right,
so
you
can
spin
up
as
a
service
as
a
registry
proxied,
but
also
as
a
CLI
right.
So
you
just
started
as
a
CLI
give
a
command,
and
it
was
just
two
trans-
the
transactions.
This
was
my
idea
to
have
also
it
usable
as
a
CLI,
but
it's
you
know
it's
not
yet
implemented.
C
E
Or
we
can
talk
offline
I
I
want
to
combine
these
two
into
one
yeah,
and
maybe
we
can
replicate
something
from
a
harbor
across
the
charts
from
Harbor
existing
Harbor
instance
and
convert
them
and
and
switch
back
to
the
existing
Harbor
I
mean
so
yeah,
that's
kind
of
yeah.
Let's
have
a
discuss
offline
about
this
yeah.
Thank
you.