►
From YouTube: CNCF Harbor's Community Zoom Meeting - Oct 20, 2021
Description
CNCF Harbor's Community Zoom Meeting
0:40 - Vadim's introduction and nomination for maintainer
19:12 - Santhosh demo to use Peritus.AI for Harbor
35:20 - General discussion
A
Right,
hello,
everyone
and
welcome
to
the
community
meeting
for
project
harvard
today
is
wednesday
october.
The
20th,
my
name
is
orland,
I'm
the
community
manager
for
harbor,
and
I'm
going
to
facilitate
that
meeting.
Please
keep
in
mind
that
that's
official
cncf
meeting
so
follow
the
call
of
conduct
or
cncf
with
that
said,.
A
B
I
would
like
to
basically
get
some
feedback
from
the
community
because
we
had
some
some
topics
in
mind
that
we
would
like
to
work
on
and
contribute
to
to
harvard,
and
I
would
like
to
get
your
feedback
and
see
you
know
where
should
we
or
where
should
we
start
first
and
also,
I
would
also
you
know
kind
of
like
to
get
some
quick
introduction
about.
You
know
myself
and
I
know
because
you
know
most
of
you
know
me
from
the
community.
B
All
the
contributors
know
me,
but
not
everybody.
So
just
you
know
give
a
quick
intro
and
also
like
the
proposal.
What
I
think
I
can
contribute.
Okay,
I
think
you
should
be
able
to
see
my
screen
yep
all
right
great.
So
yes,
hello,
everybody!
My
name
is
madim
just
quickly
about
me.
I'm
a
software
engineer.
So
I
have
a
background.
B
So
I
did
quite
a
lot
of
things.
I'm
focusing
on
satisfying
all
the
things
and
a
bit
more
about
that
later,
I'm
focusing
kubernetes.
So
companies
is
my
my
go-to
base
layer
when
I
do
things
on
this
applications,
I'm
interested
in
it's
like
the
devops
topics
and
biz
devops
topics
how
to
get
value
from
the
it
department
into
the
customer's
hands
and
how
to
make
this
visible
to
customers
that
they
can
benefit
into
this.
So
this
might
try
to
juggle
the
the
developer
side.
B
You
know
the
business
side
of
it,
I'm
also
interested
in
the
professionalism
and
personalization
I.t.
So
I'm
part
of
the
socrates
movement,
which
is
like
a
software
craftsmanship
and
try
to
you,
know,
help
our
community,
our
industry
to
be
more
professionals
or
maybe
more
more
kind
of
maybe
not
not
like
doctors,
but
seen
being
seen
more
as
engineers.
You
know
I
live
in
switzerland
and
if
I'm
not
working,
I'm
going
to
try
to
do
some
snowboarding
and
running
and
swimming
so
I'm
trying
to
be
more
active.
B
You
can
find
me
on
twitter
and
and
github
and
also
linkedin.
So
why
harvard
and
why
us
hubbard?
So
we
operate
so
my
company
is
of
operating
harbor
since
2000's
and
not,
however,
at
container
registry
since
2016,
we
do
container
registries
for
enterprises,
organizations
and
since
2020
we're
using
harbor,
because
we
think
harbor
is
the
best
container
registry
on
the
market
today,
and
we
have
this
always
upstream
philosophy,
so
everything
we
touch
and
everything
we
do.
B
We
want
to
make
a
visible
upstream,
so
we
are
not
kind
of
a
forking
harbor
and
make
stuff
on
it
and
it's
not
contributing
back.
Our
philosophy
is
really
upstream
first,
so
we
try
to
assume
everything
and
only
things
that
are
not
relevant
for
the
community,
also
all
that
relevant
to
anybody
else,
because
it's
some
operational
stuff,
we
don't
do
upstream,
but
normally
we
do
everything
upstream.
B
We
you
know
do
continuously
for
smes
and
enterprise
customers.
We
help
teams
and
organizations.
We
help
software
vendors.
You
know
using
hardware
to
distribute
their
software
as
container
images.
We
also
help
con
cloud
providers.
You
know
adapting
container
registries
such
as
ovh,
for
example,
they're
also
using
harbor
as
a
container
registry
for
their
customers,
and
we
help
so
other
other
customers
or
cloud
providers
and
tier
two
cloud
providers
adopt
harbor
into
their
infrastructure.
B
B
So
my
topic
for
today
is
oh.
My
idea
for
harbor
is
like
a
moving
harbor
into
a
container
management
solutions
or
more
something
about
container
management
rather
than
just
registry,
because
I
think
harbor
is
in
a
position
that
is
first
unique.
I
mean
there
is
no
there's
just
one
competitor
to
cover
which
is
k
in
us
in
a
sense
that
kind
of
a
the
closest
competitor
to
harbor
is
k
and-
and
we
have
the
classical
artifact
repositories
and
we
have
harper,
which
is
container
man
container
management
solutions.
B
We
know
with
applications
with
more
suitable
for
teams,
and
I
think
we,
my
idea,
would
be
to
kind
of
position.
It
more
in
in
this
area
and
make
it
outstanding
here,
and
I
have
a
few
ideas
that
we
have
been
discussing
with
customers,
and
I
would
also
like
to
discuss
those
topics
with
you,
especially.
I
would
like
to
get
you
get
your
feedback
and
what
do
you
think
has
the
most
potential
to
be
to
become
upstream
and
that
we
can
focus
on
these
topics
first
and
then
do
the
others.
B
So
I
have
a
few
topics
with
me
today.
One
is
it's
a
culture.
It's
a
deeper
integration
with
kubernetes,
so
it's
a
more
integrating
harbor
with
kubernetes,
so
interacting
it
more
with
kubernetes,
the
other
one
is
harvard
edge.
I
know
you
guys
are
working
on
that
as
well,
so
there's
there's
a
proposal
and
we
have
also
a
few
requirements
and
I
would
like-
or
I
would
be
able
to
push
this
topic
forward
and
then
the
other
topic
I
have
is
like
the
better
connectivity
and
interaction
with
other
systems.
B
So
it's
basically
more
of
an
inventic
system
that,
however,
emits
events
to
that
other
systems
can
consume
and
react
upon.
For
example,
you
know
create
a
new
project
created
well
more
than
later,
and
the
other
topic
is
contributor
infrastructure,
so
basically
helping
the
core
team
to
build
a
infrastructure
that
that
facility
facilitates
you,
know,
merging
pull
requests
or
accepting
pull
requests
on
verifying
pull
requests,
having
a
better
infrastructure
or
where
you
know
those
tests
can
be
performed.
B
So
deeper
integration
companies
means
like
exchange
image
data
with
kubernetes
I
mean
there
is.
There
is
a
few
topics
here.
I
I
didn't
write
that
much,
but
it's
basically
about
how
we
can
make
harbor
interact
better
with
kubernetes.
So
it's
not
about
making
harvard
dependent
on
kubernetes
or
making
harbor.
You
know,
you
know
kind
of
yeah
dependent
companies
but
more
interactive
with
kubernetes.
B
This
is
already
a
community
project
and
I
think
there
is
no
need
to
to
do
this
further,
but
other
ideas
that
that
would
make
sense
is
basically
kind
of
a
I'm
integrating
with
more
you
user
management
system,
for
example,
or
the
the
vulnerability
scanner
we
have
already
in
harbor
the
option
that
if
there
is
the
image
vulnerable,
you
cannot,
you
cannot
download
this,
and
this
would
might
be
could
be
improved
a
bit
further.
So
my
roughly
idea
is
like
having
a
more
deeper
integration
companies.
B
I
have
a
few
ideas
so
that
I
don't
have
my
notebook
today,
because
I
made
it
over
the
last
weeks.
I
made
a
few
notes.
What
can
be
integrated,
but
I
can
I
can
give
you
the
general
theme,
and
I
would
outline
this
in
the
in
the
community
proposal.
So
my
goal
is
to
write
a
community
proposal
on
the
I
would
say,
top
priority
topic
here.
B
The
other
area
of
where
I
could
contribute
to
help
is
the
harbor
and
edge,
and
there
is
two
options
I
mean
in
the
community.
There
was
already
discussions
called
harbor
satellite
concept,
and
here
we
have
two
options.
One
is
called
harbored
edge,
so
we
have
a
small,
tiny
harbor
instance
running
on
kubernetes
cluster
and
edge
or
in
different
environments
that
are
maybe
not
connected.
B
To
that
are
connected
to
the
main
harbor
instance,
so
there
will
be
one
major.
There
will
be
one
main
main
harbor
instance
and
multiple
edge
devices
satellites
that
are
connected
to
the
main
instance,
and
there
would
image
replication
happen
to
those
various
instances,
but
you
would
only
centrally
manage
it
and
the
other
concept
is
basically
have
a
global
harbor
where
you
have
a
master
to
master
application.
B
So
we
have
a
harbor
instance,
for
example
in
the
us
and
one
in
in
in
asia,
and
those
instances
would
replicate
the
images
but
also
the
users
and
projects,
and
things
like
this.
So
there
are
two
aspects
here
and
you
know
we
need
to
find
a
way
which
we
think
is
the
best
for
for
the
for
the
product,
and
I
would
like
to
you
know,
outline
a
few
things
here
and
and
see
you
know
where
to
move
forward,
and
I
would
you
know,
create
a
proposal
for
that.
B
The
the
third
topic
is
the
connectivity
and
interaction
with
other
systems,
so
the
the
roughly
the
idea
is
to
place
a
harbor
into
city
into
the
city,
contain
a
continuous
delivery
pipeline
more
and
make
it
a
central
load
roll
there,
and
the
idea
is
how
to
accomplish.
This
is
when
harper
would
emit
events
regarding
the
state
changes,
so
we
have
already
some
some
web
hooks
and
they're,
mostly,
I
would
say,
designed
in
a
way
to
for
for
slack
notification.
B
Things
like
this,
so
it's
kind
of
has
has
this
touch
of
it,
but
you
know
like
if
new
images
pushed
or
new
vulnerability
is
discovered
for
existing
image.
So
if
we
do
continuously
scan,
we
could
emit
emit
events,
but
there
are
also
events
that
could
be
emitted
for
new
users,
new
projects
that
are
created
or
even
for
images
to
get
pulled
or
pushed.
So
those
events
will
allow
to
integrate
harbor
better
into
the
cd
pipeline,
and
this
integrations
will
make
yeah
make
it
possible
to
develop
a
more
flexible
deployment
pipelines
for
users.
B
Solution
for
harbor,
as
well
or
potential
option
that
could
be
implemented
for
harbor
and
the
last
point
is
contributor.
Infrastructures
is
basically
having
some
sort
of
an
infrastructure
that
that
will
help
the
core
team
and
harbor
to
you,
know,
react
and
validate
prs
faster.
So
there
are
quite
a
few
pr's
that
are
still
open,
and
I
I
see
there's
a
problem,
maybe
in
validating
those
pr's
and
you
know
getting
focus
on
them
because
you
know
there's
so
much
so
much
going
on
and
I
could
help
and
you
know,
build
an
infrastructure.
B
So
we
have
to
know
how
we
have
the
people
to
do
this
and,
for
example,
one
option
would
be
to
create
a
new
harbor
instance.
Every
time
there
is
a
pr
so
that
you
can
just
log
in
and
verify
the
pr
and
then
create
a
test
instance.
So,
for
example,
the
user
that
created
the
pr
is
is
capable
of
you
know,
replicate
setting
up
and
a
test
case
that
that
can
be
validated
easily,
and
there
is
also
this
option
from
from
microsoft,
which
called
open
source
and
microsoft.
B
So
this
is
kind
of
a
the
areas
that
I
would
like
to
focus
on
and
help
contribute,
and
for
me,
the
important
part
is:
what
do
you
think
is
the
most
valuable
part
for
the
community
and
for
harbor
that
we
should
detect
go
first,
so
just
you
know,
use
the
chat
and
give
give
it
a
number.
What
do
you
think
would
be
the
the
best
option
to
start
working,
because
we
cannot
do
everything
at
once.
You
know
it's
always
like
this.
B
We
need
to
prioritize
and
start
with
one
thing
and
then
move
on
to
the
next
one.
So
what
would
you?
What
would
you
think?
Sorry,
sorry,
sorry,
what
would
you
think
would
be
the
best
option
to
to
create
a
community
proposal
so
just
give
one
two
three
or
four
in
the
chat
and
then
we'll
see
how
much
votes
we
do
have.
B
B
I
think,
for
example,
this
currently
it's
harvard
edge,
so
I
would
create
a
proposal
for
harvard
edge
and
then
I
will,
you
know,
write
down
my
thinking
how
this
might
look
like
and
also
open
the
proposal
if
it
should
be
harbor,
global
or
harvard
edge
satellite
concept
and
see
what
what
are
the
votes
on
on
those
topics
and
what?
What
are
the
feedbacks
and
and
then
outline
this.
And
if
we
have
the
agreement
on
on
the
proposal,
we
will
then
maybe
start
with
integrating.
A
Yeah,
because
that
was
the
prerequisite
to
start
actual
working
group
for
edge,
so
we
can
facilitate
the
effort
towards
edge
so
I'll.
Take
a
note
and
then
speak
with
him
if
he
can
drive
this
one
from
project
management,
point
of
view
and
somehow
combine
the
two
or
you
can
contribute
from
with
your
ideas,
everyone
with
their
ideas
to
the
initial
proposal
that
alex
worked
on.
Okay,
okay,
yeah.
B
Exactly
so,
this
is
the
the
the
point
here.
So
if
there
is
already
something
I
would
definitely
contribute,
I'm
I
was
not
kind
of
don't
want
to
start
from
spreadsheet
or
something
it
will
be
good.
If,
if
not,
we
can,
you
know,
start
the
discussion
and
and
start
a
contribution,
yeah
yeah.
So
this
is
like
a
bit
about
what
I
think
I
could
contribute
to
to
the
project.
You
know
I
would
like,
of
course,
be
more
more
involved
in
the
community.
Are
you
I
think
you
saw
me
already?
B
A
few
of
you
saw
me
already
on
the
on
the
slack
channel,
where
I
tried
to
help
people
with
a
harbor
and
installing
an
adapting
harbor,
and
you
know
I
could
also
write
some
documentation
on
on
the
on
harbor
on
the
project.
However-
or
you
know,
create
some
content
or
blog
posts
on
how
to
use
hardware,
because
we
have
also
some
some
topics
in
there,
and
I
could
also
share
this
with
the
community
and
also
share
those
this
topic
on
the
harbour
block.
B
For
example,
we
have
also
a
few
a
few
content,
for
example,
especially
how
to
integrate
harbor
with
kubernetes.
You
know
and
how
to
do
those
topics
that
would
help
and
also
writing
documentation
or
improving
documentation,
and
we
have
a
lot
of
documentation
regarding
integrating
of
hardware
with
open
id
connect
providers
like
google
azure
alt
zero.
What
else?
So
we
have
also
this
this
quite
a
few
of
this
topic
and
we
could
help
and
and
move
them
to
harvard
yeah.
So
that's
it
from
my
side.
I
have
no,
no
other.
You
know
other
topics.
A
About
the
the
documentation
part
we're
about
to
kick
off
doc's
working
group
when
abby
is
back
he's
officially
driving
this
one,
so
that
can
be
part
of
that
effort
combined
and
one
of
your
points
on
on
your
slides.
Were
you
to
be
become
part
of
the
maintenance,
maintenance
team
and
actually
that's
the
topic
that
would
be
yeah
yeah?
Actually,
that's
the
topic
that
I
wanted
to
discuss
with
the
folks
on
the
call.
A
I
would
like
to
nominate
with
him
for
maintainer
and
I'm
gonna
open
the
pr
by
the
the
community
repository,
so
all
the
maintainers.
I
would
like
to
ask
them
can
lead
to
vote
on
this
one.
You
know
the
team's
work
and
his
engagement
and
commitment
to
the
community.
So
please
when,
when
the
pr
is
up,
please
vote
for
writing.
A
D
Yeah,
so
thanks
for
the,
unlike
with
him,
I
have
no
slides,
I'm
just
going
to
do
a
straight
out
demonstration.
First,
let
me
introduce
myself.
I
am
santor
srinivasan,
I'm
based
out
of
the
bay
area,
it's
6
20
in
the
morning
here
for
me.
So
when
you
were
talking
about
3
30
a.m,
I
was
thinking
okay.
I.
E
D
D
We
are
already
working
with
cncf's
argo
project
and
we
are
already
in
touch
with.
We
had
a
talk
with
chris,
the
cto
of
cncf
and
the
work
we
are
doing
with
our
go.
We
want
to
bring
it
to
the
rest
of
cncf
free
of
cost
to
all
the
open
source
communities.
So
it's
a
no
charge,
nothing.
You
can
use
it
forever.
Today,
I'm
going
to
be
demonstrating
the
product
as
it
is
installed
in
cncf,
and
you
can
ask
me
questions
as
we
go
along
right.
E
D
It
on
chat
or
directly
ask
me
works
whatever
way.
Just
a
little
bit
of
background
about
myself,
I've
been
in
open
source
for
a
long
time
as
one
of
the
early
engineers
on
apache
hadoop
ecosystem.
I
am
a
maintainer,
slash,
committer
on
apache
pig
part
of
the
project
management
committee.
It's
a
parallel
universe
to
cncf,
I'm
fairly
new
to
cncf
for
about
a
year
year
and
a
half
in
on
the
end
user
group,
mostly
not
on
the
maintainer
stuff,
but
I've
been
working
with
and
using
open
source
for
a
long
time.
D
D
So
I'll
walk
you
through
quickly
what
I
have
done
as
work.
So
what
you're
looking
at
is
my
slack
desktop
client
on
my
mac.
This
is
a
cncf
workspace
and
we
have
an
assistant.
It's
called
pareto
system
for
slack,
so
if
you're
already
engaging
with
the
community,
what
it
essentially
means
is
you're
trying
to
find
an
answer
to
a
question.
Typically,
you
use
google,
which
is
pretty
much
the
entire
world,
and
google
is
the
best
search
engine
out
there
right.
No
doubt
there
are
alternatives.
D
D
So
you
usually
try
about
two
to
five
keywords
and
you
try
to
tweak
it
to
get
to
an
answer
and
google
has
a
cap
on
32
keywords.
So
you
can't
ask
long-form
questions
on
google
and
if
you
look
for
technical
questions
for
forum
oriented
answers,
it
either
takes
you
to
stack
overflow
or
the
equivalent
q
and
a
platform,
for
example
on
kubernetes.
It
could
be
the
discourse
or
discuss
community
on
kubernetes.
D
It
has
no
insight
into
all
the
community
activity
happening
on
slack.
Google
has
no
insight
because
it's
not
available
publicly.
So
if
you're
here,
you
already
have
gone
past
google,
so
that's
the
top
level
thing
you're
looking
for
help
with
answers
and
also
maybe
help
with
some
expertise
from
the
community.
D
Our
assistant
for
slack,
is
installed.
It's
enabled
for
the
argo
projects.
It
can
be
invoked
to
invoke
two
ways
if
it
is
enabled
on
a
channel
just
like
how
you
direct
message,
somebody
on
a
channel
but
doing
that
you
can
do
that
or
you
can
do
direct
message.
What
I'm
going
to
demonstrate
today
is
a
direct
message
with
the
bot,
because
I
don't
want
to
pollute
the
channel
conversations
I'll
pick.
Some
live
questions
to
show
how
it
works.
D
First
I'll
walk
you
through
how
it
is
then
I'll
do
a
quick
one,
because
it
takes
less
than
a
few
seconds.
So
long-form
questions
right,
you're
all
used
to
it.
There
will
be
a
text,
some
config
code,
some
more
text
so
on
and
so
forth.
And
then
you
have
a
threaded
response
for
people
who
try
to
chime
in
and
that's
when
those
questions
come
in,
what
we
do
is
we
try
to
extract
the
most
relevant
content
from
it
and
bring
answers
to
this
question
from
trusted.
D
So,
for
example,
in
this
question,
I
think
I
picked
this
up
from
one
of
the
helm
channels
or
something
when
the
question
comes
in
you
can
see,
there
is
an
answer
coming
a
recommendation
coming
from
documentation,
one
coming
from
stack,
overflow
and
so
on
and
so
forth.
With
slack,
we
try
to
be
concise
about
five.
We
are
all
constantly
deploying
new
features
to
make
it
more
usable.
We
also
say
here
are
some
experts
that
we
know
based
on
every
single
activity
you
have
had
for
this
topic
in
our
universe.
D
It
could
be
a
stack
overflow,
it
could
be
on
github.
It
could
be
on
your
own
slack
channels
right.
So
what
I'll
do
as
a
quick
demo
to
say
how
quickly
this
works
is
I'll.
Just
pick
one
of
the
questions
here
like
this
came
in
right
before
the
meeting,
so
I
can
take.
Maybe
this
one
here
I'll
just
pick
one.
So
just
you
can
get
a
flavor
of
it.
You
can
directly
ask
it
in
the
channel
or
here
I'll,
just
show
you
how
it
can
work.
D
I'm
literally
copy
pasting,
the
question
here
and
saying:
hey:
can
I
create
an
application?
You
get
a
response
almost
instantaneously
as
a
threaded
response,
and
it
will
show
you,
for
example,
here
it
is
self-referencing
the
question
that
was
asked
before
if
it
was
all
already
on
stack,
you
cannot
get
this
on
google
period
and
it
is
going
across
multiple
sources.
D
For
example,
there
is
one
github
issue
here:
another
one
is
from
a
documentation
source
and
also
we
show
some
of
the
folks
who
are
most
knowledgeable
about
this
specific
question
in
this
slack
channel,
because
this
is
an
argo
question
coming
from
slack.
So
this
is
samuel.
I
don't
know
who
samuel
is.
This
is
all
our
machine
learning
models
figuring
out
based
on
interactions?
D
D
D
Don't
get
a
question
I'll
show
you
quickly
how
it
is
non-intrusive,
it
doesn't
respond
unless
it
is
spoken
to
and
we
have
a
feature
called
nudge
which
will
it
is
about
to
be
deployed.
I'll
just
show
you
the
version
of
it
on
my
workspace.
So
here
is
the
diversion
where
I've
been
asking
questions
before
the
call,
and
if
you
have
asked
a
question,
haven't
received
a
response
in
a
finite
amount
of
time,
one
two:
three
five
hours,
we
just
nudged
them
saying
hey
by
the
way.
D
There
are
some
questions
here
sends
it
only
to
the
original
poster.
No
one
else
gets
it
and
they
can
look
at
the
recommendation
right,
so
it
just
pops
up.
They
can
take
a
look
if
they
feel
comfortable.
They
think
it's
good.
They
can
post
it
for
everyone
else
to
take
up
a
password,
so
highly
non-intrusive,
nothing
comes
in
the
way
of
the
regular
flow.
D
D
They
never
got
a
response
and
the
question
the
57
percent
of
questions
that
got
a
response.
If
you
did
not
get
a
response
in
one
hour,
there
is
a
71
chance
that
you
will
never
get
a
response,
so
it's
highly
time
based
because
there
are
so
many
slack
messages
that
if
it
goes
past
your
scroll
you're
not
going
back
to
it
right.
D
So
this
is
a
quick
demo
of
the
slack
assistant.
So
what
we
do
to
complement
this
thing
is
I'll.
Just
show
you
the
other
analytical
part
about
it,
so
we
also
do
aggregate
analysis
of
your
conversations
across
all
channels.
So
this
is
the
argo
one.
We
call
this
hotspots
and
argo
is
organized
around
cd
workflow.
There
are
five
or
six
different
things.
D
The
first
level
organization
is
based
on
how
the
community
and
the
projects
are
organized.
The
second
level
organization
is
all
machine
learning
base.
So
these
are
topics
picked
up
by
machine
learning,
algorithms,
and
I'm
going
to
pick
the
topic
here
and
what
it
shows
you
is
who
are
asking
questions
around
these
topics
who
is
responding
and
who
are
the
experts
who
should
be
responding
to
move
the
topic
along?
D
D
We
say
how
how
engage
the
community
is,
and
we
also
give
you
the
friction
score
or
the
frustration
score
for
that
right,
so
why?
Why
is
this
important?
Because
now
it
helps
the
community
figure
out
hey?
Is
there
a
bug
that
needs
to
be
prioritized
like?
I
saw
vadim
asking
for
features,
draw
analogy
of
that
for
features
bugs
missing
documentation
or
something
that
is
broken
when
a
release
goes
out,
that
you
have
to
jump
on
it
and
immediately
do
a
1.2
release
to
unblock
the
community,
so
this
is
completely
based
on
the
conversations
happening.
D
D
I'll
take
any
questions,
so
there
are
many
more
things
that
I
could
show,
but
just
wanted
to
highlight
last
thing:
I'll
leave
you
guys
with
is
just
to
show
you
the
data
that
we
collect,
for
example.
This
is
what
we
did
with
argo.
I
might
have
harvard
data.
It
could
be
dated
just
a
little
patience
with
me,
so
we
have
to
take
the
entire
universe
of
conversations,
because
not
all
communities
started
at
the
same
time
and
they
are
not
active
at
the
same
time.
D
E
D
We
also
look
at
if
there
is
q
and
a
based
things.
What
how
engage
the
community
is,
how
quickly
you're
responding
to
questions
so
on
and
so
forth,
and
we
also
grade
the
expertise
of
people
and
we
can
show
who
are
the
experts
across
communities
to
highlight
not
only
people
who
are
in
common,
but
also
entire
expertise
across
all
of
these
communities.
D
So
these
are
tools
that
will
help
you
just
benchmark
yourself
within
the
overall
ecosystem,
learn
best
practices.
You
can
talk
to
your
counterpart
projects
and
if
you
have
people
in
common
you
can,
then
you
know
talk
to
them
to
find
out
what
are
good
things
that
you
can
improve
on
so
and
so
forth.
I
might
have
harvard
here,
I'm
not
sure,
because
this
is
a
list
of
all
the
projects
that
we
had
grabbed
where
there
is
hardwood
here.
Okay.
D
So
this
is
the
hardware
stuff
that
we
could
pull
in
publicly
till
beginning
this
up
until
april,
then
yeah
right.
So
this
may
make
more
sense
to
you
guys,
because
I
have
not
spent
too
much
time
on.
It
looks
like
hardware
activity.
I
don't
have
access
to
your
slack
messages.
It's
mostly
centered
around
github
and
very
little
activity
on
public
q,
a
forums.
A
D
Okay,
so
our
assistant
is
already
installed
there.
I
can
work
with
you
to
just
get
it
enabled
on
the
channels
where
you
want
it
enabled
and
after
that
we
just
need
to
prepare
our
data
to
make
sure
you're
getting
recommendations
and
the
counterpart
intelligence
that
you're,
seeing
here
the
top
level
stuff
yeah
so
little
to
no
work.
At
your
end,
it's
just
as
simple
as
enabling
or
installing
a
slack
app
on
your
workspace.
That's
it
everything
else.
We
take
care
of
it.
A
D
Yeah
no
excellent
question:
we
don't
respond
to
a
question
unless
somebody
asks
us
directly
and
the
administrative
features
that
I
showed
will
just
to
highlight.
I'll
just
highlight
why
it
is
an
admin
feature
is
if
you
go
to
home
and
you
go
to
settings
here
so
this
this
is
called
a
nudge
feature.
So
in
case
you
want
to
nudge
people
to
look
at
the
assistant's
recommendation.
D
B
Interesting,
I
I
like
it.
I
have
a
follow-up
question,
so
the
the
data
sources
that
you
can
connect
to
is
it
is
it
limited
or
can
we
also
maybe
connect
it
to
the
documentation
of
hardware
or
to
the
to
the
community,
a
community
chat
and
github?
So
there's
a
concurrently,
a
community
not
chat,
but
community
forum
kind
of
thing.
D
Yeah
so
today,
yeah
good
question,
so
the
resources
that
we
support
today
out
of
the
box
are
github
stack,
overflow,
slack
discourse,
which
is
where
kubernetes
is
how
the
lithium
sales
force
we
just
looked
at.
Where
are
all
the
public
communities
being
hosted
by
folks,
and
we
have
built
connectors
to
get
data
from
those
sources
and.
D
D
D
B
E
D
A
C
C
A
C
C
I
I
I
just
add
the
keyword
to
the
release
now
the
about
2.4,
the
anchor
feature
of
this
release
is
distributed
tracing,
that
is,
the
the
implementation
is
based
on
the
open
time
matrix.
So
you
can
set
up
a
trace
service
and
then
hardware
can
push
the
some
information
to
that.
So
you
can
trace
that
requires
the
time
response
things
about
that
and
it
it's
helpful
for
the
debugging
about
the
troubleshooting.
C
C
C
A
E
I
think
for
the
announcement
for
the
new
release,
usually
we
we
will
do
that
after
after
the
of
the
new
release
is
out
right
and
the
either
alex
or
like
owning
or
jonas,
you
can
have
to
post
a
blog
for
the
new
release
in
it
in
slide
or
in
t
right
advertising.
The
synthetic
block,
something
like
that.
A
E
E
C
The
job
service
depends
on
the
radius
data
to
schedule
the
job
theoretical
job,
but
we
found
something
may
crop
in
the
really
snow
so
that
the
job
service
cannot
run
the
priority
job
and
the
radius
is
actually
the
cache
of
the
hardware.
We
should
not
store
any
persistent
data
in
the
video
so
so
that
we
take
this
on
the
when
you
launch
harbor,
the
job
service
will
sync
the
scheduled
data
from
database
to
redis.
So
you,
when
user,
do
the
migration
of
harbor.
B
B
E
A
E
A
E
E
B
But
do
you
think
it?
It
makes
sense
for
harbor
to
do
to
do
backup,
because
I
mean,
if
we
think
about
backing
up,
we
have
a
postgres
database.
Now
redis
is
out
of
scope
because
reddis
is
just
a
cache,
so
we
have
a
postpress
database
and
postplace
is
you
know,
there's
tons
of
solutions
there?
Maybe
we
should
just
on
a
high
level
documented
how
backup
harbor
and
on
the
on
the
s3
level,
we
have
sorry
on
the
on
the
image
storage
level.
We
have.
B
S3
and
we
have
maybe
file
system
and
and
all
the
and
you
know,
other
object
stores
and
those
there
are
also
solutions
there
for
file
storage.
Backup
is
also
solutions.
I
think
it
may
be
just
give
the
users
a
guideline
how
to
do
the
backup
and
but
you
know,
I'm
not
sure
if
it
would
be
make
sense
to
provide
a
solution
in
in
harbor
other
than
documenting
it.
C
So
now
we
actually
are
already
working
on
that,
and
you
may
know
that
the
maintainer
also
from
vmware,
so
so
that
we
have
some
discussion
about
that
team
and
we
will
do
some
integration
with
valero.
I
mean
humber,
will
valerie
and
will
eventually
provide
a
solution
for
that
to
have
the
user
in
the
committees
to
backup
and
restore
hardware
data
like
storage.
D
I
can
just
share
the
end
user
perspective.
I
was
on
one
of
those
office
hours.
I
think
all
the
maintainers
patiently
answered
all
questions
that
came
along,
which
is
pretty
awesome.
That's
how
I
got
introduced
to
how
to
attend
this
community
meeting
and
stephen
zaw
was
the
person
I
was
chatting
with
on
the
cncf
slack
to
get
here
and
other
observation
I
have
is
mostly
from
time
zone
perspective.
E
D
It
may
not
have
been
beneficial
for
a
lot
of
attendees
to
come
in
if
they
had
a
long
day
they
may
have
dropped
off.
Maybe
that
gets
solved
when
the
next
kubecon
is
held
in
europe.
So
some
of
the
other
time
zone
issues
may
get
a
lot,
but
it
was
a
really
good
session.
They
answered
every
question
that
came
along
the
way
here.