►
From YouTube: CNCF Harbor's Community Zoom Meeting - March 9, 2022
Description
CNCF Harbor's Community Zoom Meeting
A
Hello,
everyone
welcome
to
harvard
community
meeting
today
is
march,
the
9th
and
that's
official
cncf
meeting.
So
please
follow
the
code
of
conduct
for
cncf
with
that
I've
pasted
in
the
chat
the
link
to
the
howard
community
meetings
notes.
So
if
you
have
any
anything
to
add
to
to
it
like
the
topics
that
you
want
to
discuss
today
feel
free
to
do
that.
My
name
is
louis
vasilev,
as
it's
written
on
my
video
stream
without
I'm
gonna
share
the
the
meeting
notes
from
today
for
today
give
me
a
sec.
A
A
Okay,
I'm
not
sure
who
added
this
one.
D
E
D
Okay,
hello,
I'm
eva
from
tencent
and
I'm
glad
to
representation
our
proposed
here
and
we
propose
to
adapt
multiple
type
of
database
for
harbor,
and
the
first
step
is
to
support
my
circle
in
mariadb
and
there
is
an
abstract
door.
Layer
in
harbor
and
different
databases
can
help
their
drivers
to
implement
the
interface
and
we
want
to
add
my
circuit
driver
and,
let's
see
the
background.
D
So
in
order
to
keep
consistency
with
clear
and
reduced
maintenance
difficulties,
harbor
unified
database
using
pg
and
since
hubble
version,
2
used
trivia
as
default
scanner
instead
of
clear.
So
there
is
no
strongly
requirement
to
use
pg
anymore.
Therefore,
it
is
possible
to
adapt
different
kinds
of
database
now
and
our
goals
are
as
follows.
D
D
Also,
if
others
interesting
it's
it's
good
to
support
other
data
in
the
future
and
second
implement
mario
db
and
my
circle
of
operator
for
internal
database
case.
At
present,
we
expected
to
support
support
external
database
usage
and
if
internal
database
user
usage
is
supported
depending
on
the
future
requirements,
if
need.
D
D
And
I'll
show
the
migrations
directory
architecture
later
then
they
need
to
success
and
when
their
work
they
need
to
provide
service
need
to
interact
with
the
database
and
job
service.
Core
exporter
will
use
the
certain
driver
type
to
access
the
database,
as
for
notary
itself
originally
supports
my
circle
usage.
We
only
need
to
modify
the
configs
to
provide
the
right
database
config.
D
D
D
And
migrate
db
with
certain
circle
file,
pg
circle
file,
founder
poster
brass
and
my
circle
circle
file
on
the
db
underneath
under
my
circle
and
here's,
the
directory
architecture
of
migrations
file
names
are
the
same
under
my
circle
and
the
postgres
that
falcon
10
may
be
different
because
of
different
grammars.
Some
some
grammars
are
different.
D
D
D
D
We
have
done
some
tests
for
circle,
compact
compatibility
and
he
released
some
support
in
incompatible
points.
First,
pg
use
trigger
to
update
the
files
of
update
time
and
instead
of
3
trigger
my
circle,
use
default
and
update
use.
This
way,
it's
okay-
and
this
is
the
one
point
and
next
serial
type
in
my
circle
is
big
inch
on
lined
and
this
type
is
not
supported
to
be
a
foreign
key.
D
D
D
D
D
D
D
D
F
G
E
D
D
D
D
The
basic
functions
looks
okay,
delete,
project,
delete
and
the
time
will
be
only
modified
as
a
necessary
code
and
to
verify
that
my
circle
is
available
and
we'll
refine
the
code
later,
and
this
proposal
still
have
many
works
to
do
so.
If
you
are
interested
in
this
proposal,
welcome
to
join
our
work
groups-
okay,
all
right,
please.
I
finished
my
presentation
thanks.
C
We
have
several
a
question
about
your
proposal
assessed
for
the
presentation.
I
didn't
see
a
very
clear
part
in
the
this
proposal
to
support
natalie
all
of
things.
Naturally,.
E
C
Yeah,
so
we
have
to
add
this
part
to
this
proposal
to
support.
How
do
we
support
notary
to
support
other
kind
of
database?
C
Okay,
oh
yeah,
the
other
one
is
the
yes
so
and
as
you
list
here,
there
are
several
other
cq
are
not
passed
in
this
table
right.
So.
C
D
C
D
C
And
me
have
introduced
some
make
fireworks
and
you
have
to
add
some.
My
secret.
C
We
have
to
add
this
part
and
for
the
compatibility
test,
since
we
use
some
specific
feature
for
in
the
cq
files,
so
we
how
to
handle
this
part
we
we
should
add
this
in
in
your
proposal.
This
is
my.
H
C
C
But
in
my
opinion,
we
have
to
provide
a
framework
to
have
user
to
add
some
new
implementation
to
support
other
kind
of
migration
like
if
I
implement
another
kind
of
driver
to
support
like
my
mongodb
yeah.
C
How
can
we,
how
can
I
migrate
my
data
from
post
requests
to
mongodb
and
how
can
I
implement
a
new
adapter
such
as
our
plugin
in
your
migration
tool?
So
for
the
part
it
should
be
a
framework,
I
think
not
just
simply
already.
D
G
Be
generous,
I
I
I
understand
what
you
said.
You
mean
we
need
to
write
this
migration
tool,
abs,
abstractly.
We
need
to
support
other
database
in
the
future.
We
will.
We
will
add
this
part
in
the
proposal.
C
Now
we
should
have
more
details
on
the
migration
partner.
How
can
they
use.
H
H
One
yes
yeah
just
related
to
the
migration
part,
so
I
I
also
had
a
thought
that
we
should
be
a
little
like
since
we
use
a
database
driver
abstraction
which
is
gobigo.
I
think
it's
a
bigo
framework
that
is
used.
We
need
to
be
aware,
like
we,
we
need
to
benchmark.
What
will
be
is
the
behavior
of
bigo,
like
whatever
apis
we
use
of
bigo?
Are
they?
Are
they
like?
H
You
know
it
may
not
be
sent
to
send
similar,
but
there
should
be
some
level
of
similarity,
because
it
should
not
happen
that
bigo
implementation
creates
a
performance
problem
when
you
start
migrating
the
data
from
postgre
to
mysql,
because
we
do
not
know
if
the
implementation
has
used
something
specific
to
postgrade,
to
speed
up
some
executions
and
we
are
just
invited.
We
are
just
using
that
feature
without
knowing
it.
H
So,
in
my
opinion,
in
addition
to
the
migration
abstraction,
as
yan
mentioned,
which
is
which
is
really
important,
we
also
should
make
sure
that
the
the
sdk
that
we
are
using
to
interact
with
the
database
right
now
is
going
to
perform
comparably
across
both
the
databases
and
it
does
not
induce
some,
like
you
know,
lags
or
performance
penalties
which
are
hard
to
track.
So
I
think
that's
important.
H
I
guess,
since
we
are
moving
across
databases-
and
it's
not
like
a
general
driver
like
bigo,
will
definitely
use
some
hints
to
optimize
some
types
of
some
types
of
executions
or
other.
So
we
need
to
check
that
out
too.
I
guess.
G
You
mean
we
use
big
or
have
some
performance
issue
right.
Yes,.
H
That's
right
like
like
it
may
be.
Speed
like
the
same.
The
same
functionality
will
be
speedy
will
be
executing
very
well
on
postgre,
but
the
moment
you
migrate
to
your
moment
you
migrate
to
mysql
or
mario
db,
possibly
like
you
know,
you
may
see
a
performance
hit
and
it
could
simply
be
because
the
underlying
big
implementation
is
doing
something
wrong,
and-
and
I
call
this
out
because
I
have
done
some
database
migration
projects-
I
have
seen
this
happen
in
the
java
world.
H
E
D
C
Oh
sorry,
I
may
have
some
further
questions
about
regarding
to
the
dl
interface.
C
Do
we
have
to
implement
the
all
the
dl
interfaces
in
harbor
I
mean
for
each
manager
like
repository
type
label
and
etc,.
D
C
D
I,
in
my
opinion,
it
should
be
a
architecture
that
we
should
give
a
certain
database,
the
the
implementation,
and
that
we
can
take
a
reference
to
the
default
interface.
So
we
we
can
reuse
the
function,
reduce
the
implementation,
but
we
should
have
a
driver
to
reference
to
the
default
function.
D
B
I
have
a
suggestion,
I
think
it's
better
for
your
tools
to
discuss
offline.
B
I
think
it's
much
easier
for
us
because
yeah
about
implementation
details,
I
think,
is
a
bad
choice.
C
Yes
feel
free
to
schedule
meeting,
but
we
I
like
to
discuss
some
details
in
this
meeting
so
that
the
other
users
from
the
community
can
understand
the
proposal,
details
and
what's
going
on
on
in
the
pearl
harbor
community.
So
but
okay,
we
can
we.
I
can
just
add
some
comments
in
this
proposal
and-
and
we
can
have
discuss
on
that
so
that
others
can
see
the
communication
based
on
our
conversations.
Yeah.
C
Okay,
so
my
last
question
is
that
we
should
provide
some
guidance
for
the
developer
from
community
to
tell
them.
How
can
they
implement
a
new
database
driver?
What
should
I
do?
What
should
I
follow
and
so
that
the
others
can
follow
this
guide
to
implement
their
own
database
driver
and
contribute
back
to
harbor
like
I
want
implement
one
mongodb
driver
and
just
follow
your
gut
and
then
send
me
my
pr
on
back
to
contribute
to
harvard.
B
Yes,
I
agree
about
this.
I
have
I
want
to
discuss
some
common
things
about
this
proposal
for
databases.
Shall
we
extend
the
scope
to
too
wide?
We
will
we
need.
We
need
to
support
different
kinds
of
databases
that
others
want.
I
think,
if
we
support
too
many
databases,
it
will
be
a
big
burden
for
hubble.
To
maintain
one
of
the
things
is
your
progress
mentioned
the
migration
tool.
I
think
it
will
be
difficult
for
hubble
to
maintain
too
many
databases,
tools
and
the
logics.
C
I
think,
from
a
perspective
of
harbor,
we
just
only
maintain
post
requests,
but
if
someone
from
community
they
want
to
maintain
my
secure
mongodb
marine
db
is
good
for
us,
so
by
default,
hardware
supports
post
awards
and
all
the
maintainers
will
provide
support
help
on
that,
but
for
the
other
kind
of
database.
C
We
should
leverage
the
effort
from
community
to
support
that,
but
for
just
like
what
we
did
for
the
replication
like
the
tencent
adapter,
I
think
you
guys
could
provide
some
help
on
that
adapter
to
support
hardware
replication
from
hardware
to
tencent
network
right
yeah.
That's
my
answer
for.
B
For
your
question
from
our
from
us,
we
we
only
want
to
focus
the
marine
db
and
message
for
the
moment
and
yes,
I
will.
We
agree
that
we
should
provide
a
framework
to
have
others
to
implement
their
own
in
java,
for
layer
prefer
the
databases
but
it's
difficult
to,
for
I
suggest
by
stephen
we
we
maybe
we
will
make
a
database
whole
group,
but
I
think
it's.
If
we
didn't
control
the
scope
it
will.
B
It
will
be
very
difficult
for
the
work
groups
to
work
if
we
cover
too
many
kind
of
databases-
and
we
didn't
have
that-
that's
effort
to
do
the
work.
C
We
won't
do.
I
think
you
can
just
provide
effort
to
support
mysql
on
my
rdb
yeah.
C
For
other
kind
of
database,
I
think
if
someone
from
community
have
interest
on
that,
they
can
just
contribute
that
part.
B
Okay,
I
got
it
and
I
have
another
another
things
to
say
about
this
proposal
is
maybe,
if
we
choose
to
provide
one
kind,
a
kind
of
databases,
but
it
it
may
not
work
work
for
all
of
the
hubble
component,
for
example
the
notary.
B
If
we,
if
one
customer
wants
to
use
hubble
with
cassandra,
for
example,
but
notary
didn't
didn't,
doesn't
support
cassandra,
I
think
it
also
can
be
a
choice.
Just
if
we
make
our
choice
to
use
the
cassandra,
we
must
give
up
the
notary
use
component
just
to
use
the
basic
hardware
functions.
Do
you
think
it
is
reasonable.
C
So
my
opinion
is
not:
we
have
to
provide
all
the
functionality
for
a
new
kind
of
database,
but
the
notary
is
is
a
special
case,
and
since
we
we
have
some
discussion
or
not
with
application,
but
anyway
I
I
my
suggestion
is
that
we
should
provide
full
functionality
on
new
kind
of
database.
B
So,
if,
if
we
want
to
provide
four
functions
of
hubble
none,
if
we
want
to
support
one
database,
we
need
to
adapt
to
the
hardware
or
clear
such
candle
for
solar
parts
component
right.
C
Yes,
but
there
are
not
so
many
third-party
components
that
need
to
be
accessed
database
just
another
so
far
and
there
were
not
some
additional
third
party
not
to
be
introduced
into
harvard
to
have
access
to
database.
C
My
real
post
request,
I
don't
know
others,
sorry.
B
Okay,
so
considering
for
this,
I
think
for
the
moment
we
can
only
support
the
previously
mentioned
databases
for,
for
now,.
C
Yes,
yes,
but
if
we
duplicate
the
notary
I
mean
the
without
notary
integration.
We
can
introduce
some
other
kind
of
advice.
I
think.
B
B
We
want
to
provide
the
full
functionality
of
the
hubble.
Maybe
the
choice
of
database
is
very
limited.
C
My
personal
opinion
is
that
we
have
to
provide
function,
full
functionality
to
support
new
kind
of
database,
but
I'm
not
sure
others
of
any
or
not.
C
So
we
can,
you
can
just
put
this
item
into
your
proposal
and
then
anyone
who
reviews
the
proposal
can
give
our
answer
on
that
is
that,
okay,
for
you.
A
Great,
I
I'm
sorry,
we
have
to
run
fast.
We
have
less
than
15
minutes.
Okay,
I'm
sorry!
I
don't
want
to
put
you
off,
but
I
don't
want
to
crawl
the
hour.
So
we
agreed
that
I
can
see
the
pr
put
the
proposal
in
so
you're
gonna.
A
A
Thank
you
a
lot
for
your
demo
and
presentation
very
useful.
So
with
that,
I
think
we
can
continue
with
few
more
topics
on
the
agenda.
I'm
sorry
again
to
get
you
all,
but
we
don't
have
to
across
the
hour.
So
I
have
two
topics:
kind
of
h3.
A
Our
maintainer
truck
was
accepted
for
the
kypcon
eu
2022.,
please
I
think
young
one
you
you've
submitted
it.
Please
start
preparing.
If
you
need
my
help,
just
drop
a
line,
and
I
can
I
can
help
out
with
the
with
the
preparation
for
the
for
the
session.
I
Oh
only
that's
that's
not
true,
and
the
original
plan
is
we
target
to
release
around
may
march
11
right
yeah.
I
remember
in
last
community
meeting,
maybe
have
checked
me
to
check
the
date.
At
that
time
I
updated
in
the
stack
channel
I
it
targeted
to
maybe
march
21
since
yeah.
That's
a
monday
right,
but
currently
we
have
some
new.
I
mean
new
new
sense.
I
I
I
mean
information
from
northwest
that
when
they
can
reduce
the
patch
to
fix
gstv,
so
we
can
consume
the
upstream
patch
in
our
harbor
side,
so
that
will
also
impact
our
2.5
release.
I
think
so.
I
A
Okay,
cool
about
that,
I'm
trying
to
to
find
a
way
through
the
cstf
governance
with
the
notary
folks
if
they
can
touch
that
yeah.
Thank
you,
but
for
how
how
much
it
will
take
at
the
time
but
yeah,
okay,
abby
you!
You
have
a
topic
for
the
sick
dogs.
F
Yes,
hello.
I
just
wanted
to
plug
our
our
next
meeting,
the
sig
docs
group,
which
will
be
next
wednesday,
which
I
think
is
the
16th.
I
might
have
have
the
wrong
date,
but
it'll
be
next
wednesday
and
it
should
be
at
the
same
time
as
this
meeting.
F
We
are
having
daylight
savings
time
change
this
weekend
in
the
u.s,
so
I
might
have
put
out
the
wrong
time
so
I'll
double
check
the
date
and
the
time,
but
it
will
be
next
wednesday,
so
folks
are
interested
in
joining
and
and
helping
us
with
the
docs
or
learning
more
about
our
docs
for
the
harper
project,
we'll
be
having
a
meeting,
and
I
hope
to
see
you
all
there.
A
Right,
do
you
need
any
help
from
my
site
to
like
do
the
I
don't
know,
announcement
in
channel
or
tweet
or
something
that
we're
gonna
meet,
and
that
will
be
the
first
official
sickbox
meeting.
F
A
A
And
one
again
to
refer
back
to
the
cvs,
those
came
out
from
the
app
check
things
that
we
run
right:
yeah
yeah
and
we
okay.
How
should
I
put
this
one
if
notary
are
not
like
cooperative
enough
in
in
some
time?
What
are
our
options?
Do
we
have
any
any
other
options?
I
mean
to
like
at
least
close
our
eyes
for
this
one
and
release
like
this.
I
don't
like
this
one,
but.
A
Yeah
personally,
we
don't
have
roger
on
the
call.
Maybe
he
can
put
more
this
one,
but
I
mean
a
part
of
the
procedural.
Like
closing
our
eyes.
Do
we
have
some
technical
solutions
for
this
one?
I
think
we
discuss
once
to
fork
notary
and
fix
it
on
our
own,
which
is,
I
think
we
should
go
for
this
one
as
a
the
last
possible
solution.
A
And
in
did
you
already
submitted
a
pr
to
the
notary
with
the
fix
or
we're
just
waiting
for
them
to
fix
it
on
their
own.
C
They
already
have
someone
on
risk
pr
to
fix
that,
but
they
have
to
merge
them
and
then
raise
another
pr
based
on
that,
so
to
fix
the
another
pcbe.
But
I
I
have
already
mounted
them,
I
mean
not
for
community
or
a
maintainer.
They
have
no
idea
about
the
eda.
A
I'm
just
trying
to
figure
out
how
I
can
change
them
properly,
so
I
can
see
we
don't
have
an
open
pr
for
this
one,
at
least
from
our
side.
It's
like
yeah.
C
Yes,
yes,
the
third
one
is
up
from
offside
to
fix
the
program,
yes,
and
that
one
is
from
a
stronger
cr
to
fix
the
form
of
dependency
and
another
one.
The.
C
Return
is
the
other
one.
Can
you
switch
back
to
the
the
pianist
yeah.
A
C
The
sec,
the
six,
the
sixth
one,
this
one.
C
A
So
I'm
gonna,
I'm
gonna
talk
with
justin.
I
know
him
and
I'm
gonna
ask
him
if
he
can
provide
some
some
info
on
this
one.
C
Okay,
thank
you.
So
when
shall
we
make
a
decision
for
very
part,
do
we
have
a
enough
day
for
for
this.
C
Yeah,
whether
we
should
or
or
just
wait
for
their
response
and
eventually
we
we
should
have
a
decision
on
that
yeah.
So
I
mean.
When,
should
we
have
that
decision.
A
Okay,
that
that
depends
how
much
time
we
need
after
we
make
that
decision.
If
we're
gonna
meet
the
21st
of
march
release
date,.
C
So
we
have
already
deleted
the
lab
patch
release
for
several
weeks
and
yeah.
A
I
can
help
out
with
this
one,
but
I
I
cannot
take
that
decision.
It's
it's
not
in
my
field
at
all,
but
also
relatively
new
cvs
right.
A
Okay
but
yeah,
but
taking
into
account
that
notary
hasn't
been
released,
there's
no
new
release
for
for
notary
for
almost
a
year,
or
maybe
two.
A
Okay,
yeah,
actually
almost
yeah
three
years:
okay,
I'll,
try
to
get
into
justin
and
see
how
it
goes
and
yeah
because,
as
I
mentioned
somewhere
we're
not
the
only
project
that
relies
on
notary
heavily,
so
they
have
to
fix
that
anyway
and
yeah
yeah.
The
thing
is
that
not
very
easy:
it's
a
cnn
project,
yes,
and
they
have
some
responsibility
based
on
this
one,
so
they
have
to
do
something.
I
C
We
just
follow
their
security
pipeline
just
start
mile
to
their
security
work
group
and
to
notify
their
their
cve
in
their
project,
and
you
have
to
provide
fix
on
that,
and
this
is
the
usually
process
for
us
reporting.
A
security
issue.
A
C
I
Yeah
only
what
what
I
mean
to
report
a
security
issue
normally
will
not
file
an
issue
in
github.
You
will
follow
their
process
to
send
the
email
to
their
security
group.
To
mention
that
I
think
I
already
have
sent
an
email
to
their
security
group
and
get
a
response
from
then
so
we'll
patch
this.
But
there
is
no
ek
for
that.
A
A
I
A
Okay,
can
you
young
one,
can
you
look
me
into
that
mail
that
you
sent
so
I
can.
I
can
follow
up
from
my
own
register.
Okay,
sure
sure,
okay,.
C
I
also
passed
the
security
us
in
there.
You
can
open
that
url
to
see
some
details.
A
A
So
I
think
I
think,
a
simple
issue.
Okay,
can
you
fix
that
or
when
you
plan
to
fix
that
it
will
be
good
enough
for
now,
but
okay,
I
think
you
brought
that
to
the
security
team.
If
you
can
forward
that
mail
to
me
and
I'll
try
to
get
in
touch
with
justin,
okay,
great
okay,
with
that,
I
think
we
are
ready
with
the
agenda.
Anyone
else
or
probably
the
agenda
here-
wants
to
say
something
wants
to
share
something.
H
Hey
jan
and
team,
I
just
this
has
been
a
couple
of
weeks
ago.
H
There
are
two
reviews
and
I
have
responded
to
all
the
comments.
There
are
some
comments
like
getters
back
like
I
did.
Those
discussions
are
like
I
mean
I
think,
they're
rotating
around
the
point
of
whether
we
need
to
show
system
artifacts
to
the
user
or
not
so
yeah,
and
if
baby
has
already
responded.
H
If
you
could
take
a
comment
from
our
tian
on
respond,
it
will
be
great
so
that,
because
I
have
my
implementation-
and
I
want
to
run
this
like
as
soon
as
possible
in
the
next
couple
of
weeks,
so
just
wanted
to
make
sure
that,
like
we
don't
we
don't
get
back
again
on
a
decision
that
does
something
that
adds
some
ebook.
H
C
I
mean
I
have
already
approved
your
proposal
and
regarding
to
the
implementation,
I
think
you
can
just
firstly
to
create
an
ipad
in,
go
hub
repository
and
then
do
the
task
breakdown.
You
can
just
create
some
details
task
in
the
repositories.
So
then
we
can
use
that
issues
to
track
the
implementation,
progress.
H
Okay,
sure
sure
sure-
and
I
can
do
that
and
I'll
sync
up
with
you-
if
I,
if
I
get
blocked
somewhere,
creating
the
issue
I'll
come
with
you
so
yeah.
Thank
you.
A
Hello,
can
you
add
that
thing
to
to
the
agenda
for
today?
So
we
don't
forget
about
it
and
next
time,
if
needed,
to
discuss
again.
H
Yeah
sure
sure
I'll
add
it
to
the
jitter
today
and
I'll
I'll.
Just
put
it.
H
That
I
wanted
to
mention
so
that
you
know
the
community
knows
that
I've
started.
You
know
dabbling
with
the
code
and
getting
things
like,
but
I've
not
done
specific
tasks.
I
will.
I
will
maybe
I'll
create
those
tasks
as
young
suggested,
and
I
will
so
that
the
thing
can
be
tracked.