►
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
Hey
everyone
welcome
to
the
harvard
community
meeting,
so
I
hope
you
had
a
good
break
and
the
agenda
for
today
is
we
have
one
of
our
maintainers
we'll
be
sharing
our
progress,
we're
making
on
prometheus
integration,
and
then
I
believe
we
have
some
folks
from
susie
who
will
come
and
talk
about
their
new
private
registry
just
based
on
power
by
harbor
2.1.
A
Susie
team-
just
you
know
briefly,
introduce
yourself
and
walk
us
through
the
the
new
registry.
C
Hi
zayali,
I
don't
think
we
can
hear
you
if
you're
unmuted.
D
Good
so
yeah,
I'm
sally.
I
work
at
susie
and
I'm
here
with
some
of
my
colleagues
as
you
can
see
the
stefan
doug
flavio
and
yoshi,
and
we've
been
working
on
through
the
private
registry,
which
is
based
on
humboldt
one.
D
D
D
Things
have
been
working
pretty
smoothly
since
we
started
and
yeah
we've
also
been
able
to
make
some
contributions.
Some
enhancements
that
we'd
like
to
talk
about
and
we've
pushed
some
upstream.
We
have
some
open
issues
and
per
request
that
yeah.
I
would
be
happy
to
share
with
the
community
and
talk
about
it
a
little
bit
more.
D
So
we
have
some
security
hardening
enhancements
that
we've
made,
and
also
some
hem
test
suit,
that
we've
added,
which
we'd
yeah
we'd
like
to
talk
about.
If
everybody's
interested.
B
So,
yes,
we're
really
excited
to
to
be
here
to
have
been
using
the
harbor
products
for
quite
a
while
now,
and
we
really
like
everything
that
we
see.
So
this
is
a
big
thumbs
up
to
everyone
here
and
to
the
entire
community.
You
guys
are
doing
great
job,
so
we
we've
been,
we've
been
contributing
mostly
around
how
we
package
harbor
and
how
we
install
it
through
the
helm
chart,
and
we
have
some
improvements
that
we
integrated
in
in
this
user
private
registry
product
that
we
some
of
the
some
of
them.
B
We
have
already
pushed
as
requests
upstream,
and
we
would
really
like
to
give
them
back
to
the
community.
I
think
they
would
make
a
great
improvement
for
harbor.
Most
of
these
are
around
security
hardening,
as
I
was
saying,
and
we
have
a
list
of
pull
requests
that
target
those
topics.
Some
of
them
revolve
around
generating
passwords
installing
them
in
secrets.
B
C
B
B
B
B
Briefly,
adds
a
couple
of
options
to
values.yaml
that
allow
us
to
control
when
all
the
various,
because
there
are
quite
a
lot
of
them
in
harbor-
can
be
rotated
and
I'll
just
quickly
point
you
to
various
data
for
a
quick
overview
of
this.
So
there
are
these
rotate
certs
bull
values
that
you
can
set
to
false
to
prevent
the
certificates
from
being
automatically
rotated
everyone.
B
Every
time
you
run
hardware,
sorry
home
upgrade
for
hardware,
and
this
helps
on
scientists
there's
this
section
in
the
hardware
documentation
where
you
instruct
users
to
save
the
existing
certificates
in
order
to
prevent
this
from
happening.
This
is
a
possible
solution
for
that
as
well,
and
let
me
just
add
that
nothing
changes
in
the
default.
We
were
really
careful
about
that
in
the
default.
Behavior
that
comes
with
the
existing
hardware
hand
chart.
So
if
you
don't
set
these
values
to
different
values,
these
options
are
different
values.
B
B
B
Only
if
you
leave
this
onset
so
set
it
to
an
empty
value
or
something
like
that,
it
gets
automatically
generated
to
a
strong
password
value
and
the
other
nice
addition
to
this
is
we
don't
generate
it
every
time
you
hit
home,
upgrade
you
you
generate
it
once
and
it
gets
stored
in
secret,
and
then
it
gets
red
on
subsequent
helm
upgrade
runs
from
that
secret
and
to
do
that,
we
we're
using
a
helm
feature
called
let's
see
if
I
can
get
it
here,
look
up.
B
B
V2
and
that's
why
I
approached
this
subject
a
couple
of
weeks
ago
during
the
office
hours,
so
I
don't
know
how
to
do
this,
because
if
you,
for
instance,
if
you
look
at
the
this
pull
request,
it
does
not
pass
the
ci
on
so
this
job,
the
v.2
helm,
linter,
there's
a
bug
in
in
there
that
doesn't
prevent
that
doesn't
allow
the
lookup
feature
to
be
used.
It
will
probably
never
get
fixed
in.
Actually
I
got
official
confirmation
that
you
will.
It
will
not
be
fixed
in
helm
v2,
because
that's
out
of
support.
B
B
However,
in
order
to
do
that,
I
also
have
to
touch
the
hardware
code
and
and
the
hardware
docker
compose
configuration
files.
I
still
haven't
gotten
that
to
work
quite
well
yet
and
that's
why
this
request
is
still
in
work
in
progress.
I
will
make
some
progress
on
that
in
the
following
days.
Hopefully,
if
you
can
help
out
with
that
too,
then
thank
you.
I
really
appreciate
it,
so
these
are.
E
I
want
to
thank
you
for
your
contribution
yeah,
I
I
I
I
noticed
that
I
mean
I
recognize
you
that
you
have
submitted
quite
a
few
good
pull
requests
and
some
of
them
get
merged.
I'm
one
of
the
maintainer
of
harbor
helm.
Chart
I
and
wing
kai
are
mainly
maintaining
this
ripple
and
triggering
the
issues
and
the
pull
requests
regularly.
E
Recently
we
were
a
bit
I
mean
we
have
been
a
bit
distracted
by
the
our
you
know,
company
work,
but
I
I
think
we
can.
You
know
prioritize
your
prs
and
you
know
we
can
discuss
offline
on
the
slack
channel
in
cncf
I
mean
the
cncf
hardware
devs,
that
channel
and
generally
there
may
be
some
disagreement
generally.
We
don't
want
to
introduce
too
much
logic
in
the
chart
itself.
E
So
some
of
the
feature
you
wanna
achieve-
maybe
you
should
you
know,
move
you
know,
delegate
to
some
external
tools,
but
that's
we.
We
can
handle
them
case
by
case.
I
just
want
to
thank
you
for
your
contribution
and
yeah.
I
will
prioritize
these
prs
and
follow
up
and
review
and
discuss
with
you
on
the
slack
channel
now.
I
know
your
time.
B
Zone,
yes,
we
are
in
cd
plus
one
germany.
E
Okay,
okay,
yeah,
that's
easier
than
us,
so
yeah.
We
can
continue
discussion
on
that
and
you
know
handle
your
pr's
with
power.
B
You
can
probably
see
here
it's
it's
2
p.m,
to
15
p.m.
Okay,
where
I
am
right
now
and
thank
you
yes,
we
we
understand
that
some
of
these
changes
might
be
well
changing
the
way
that
the
hem
chart
works
and
that's
why
we
put
a
lot
of
work
into
preserving
the
default
behavior.
So
you
you
get
all
these
extra
features,
but.
C
B
Any
impact
to
upgrades,
so
that's
something
that
that
we
can
work
on.
E
Yep
thanks
for
sharing
and
yeah,
let's
discuss
offline.
F
Hello,
I
have
maybe
a
few
questions
to
stefan,
so
stefan
do
you
know
if
it
will
probably
work
if,
if
with
the
set
manager,
do
you
do
you
have
it
kind
of
tested?
If
search
manager
will
work
with
the
certificate
management
that
is
also
offered
by
a
research
manager
that
will
it
kind
of
interfere
with
your
setup
or
or
is
it
kind
of
yeah.
F
B
Those
changes
are
so
those
changes
are,
I
think
they
are
meant
to
be
used
only
with
self
generated
or
auto
generated
certificates.
So
in
that
case
you
don't
even
use
search
manager.
You
just
so
imagine
this
use
case
right.
You
want
to
deploy
hardware
quickly.
B
No,
you
don't
want
to
consult
yourself
with
certificates
or
anything
like
that.
So
you,
if
you
do,
that,
you
basically
use
this
option
to
generate
those
certificates
automatically
and
they
are
generated
automatically
by
the
hem
chart
by
the
harbor
hand.
Chart
then,
if
you
want
to
go
that
way,
and
sometimes
you
need
to
rotate
those
certificates
exactly.
F
Yes,
yeah
and
okay-
normally
I
mean
search
manager,
has
two
functionalities.
One
is
this
acme
part,
you
know,
let's
encrypt
and
the
other
one
you
can
have
your
own
self.
You
can
all
you
can.
You
can
create
your
own
ca,
you
know,
so
you
can
only
create
your
own
ca
and
assign
certificates
to
new
parts
and
new
services
when
they
come
up,
so
you
can
just
I
mean
certain
manager
can
do
this
as
well,
and
I
was
actually
more
referring
to
this
part
than
to
the
ingress
part
of
cert
manager.
That's
also
there.
B
C
B
Able
to
yeah
not
not
just
that,
but
you
want
to
maybe
not
regenerate
those
certificates
every
time
you
change
something
in
the
in
the
configuration
and
you
run
home
upgrade
because
that's
what's
happening
currently
right.
That's
that's!
What
this
idea
initially
originated,
running
helm
upgrade,
will
regenerate
those
certificates
and
there's
no
way
to
just
stop
it
right,
and
you
have
to
instruct
your
all
your
clients
to
use
those
new
certificates.
It's
it's
a
bit
of
a
nuisance,
and
this
is
a,
I
think,
is
a
nice
addition
to
that.
E
B
D
Yeah,
maybe
we
can
just
pass
it
on
to
flavio
since
he's
the
one
who's
worked
on
it.
If
we
have
time
that
is
we'd
like
to
talk
about
the
handcast
feature
that
we've
added
there's
only
an
open
issue
for
it,
we
haven't
submitted
the
pull
request
yet
but
yeah.
If
there's
time,
we
can
elaborate
a
bit
on
that
and
take
up
some
more
questions.
If
there
are.
B
A
Oh,
it's
fine!
Well,
you
know
we,
we
wanted
you
to
come
and
maybe
share
some
progress
on
the
newly
released
private
registry
that
you
have
and
sort
of
the
the
challenges
and
the
problems
you
came
across
while
while
using
harbor,
what
can
we
do
better?
Why
did
you
choose
harbor
things
like
that?
But
if
you
want
to
share
specific
improvements
like
the
pr's,
you
talked
about,
and
that's
that's
fine
as
well.
So
it's
your
time.
E
G
Okay,
so
I
will
I'll
try
to
be
briefly
talking
about
this
home
test.
So
as
we
deliver
our
private
registry
using
ham
chart,
we
went
to
the
way
to
test
it
and
we
implemented
that
using
the
helm
test
feature
and
the
way
that
it
works
is
just
running
the
api
tests
that
are
currently
run
only
on
the
hardware
project.
G
These
sets
are
run
by
hci
and
the
way
that
we
implemented.
That
is,
that
we
just
run
helm
test
and
it
will
create
a
pod
and
run
the
same
test,
but
just
using
the
helm
chart.
So
we
want
to
contribute
that
also
to
the
hemp
chart
project
but
well.
The
big.
G
The
most
complex
part
is
that
we
do
not
have
an
image
with
the
test
that
I
release
it
release
it
together
with
the
other
image
and.
G
G
Yes,
sorry,
so
the
the
the
main
problem
with
that
contribution
is
that
we
do
not
have
a
image
with
the
tests.
So
when
you.
G
Yeah,
we
need
an
image
having
the
test
that
so
we
can
run
the
test
when
we
read
the
hem
test
common
and
I
have
opened
an
open
issue
for
that
on
the
hardware
project
and
so
far
we
didn't
get
much
feedback
on
it.
I
can
link
it
here
on
the
chat.
Yes,.
D
Please
thanks
love
you
to
answer
your
question
from
before,
so
as
to
why
we
chose
harper.
D
There
were
a
few
things
that
really
stood
out
because
it
is
a
cncf
project
and
we
saw
that
there
is
a
good
community
backing
the
project
and
also
some
of
the
features
that
were
important
for
us
was
that
it
was
oci
compliant
and
the
oidc
integration
was
something
we
were
looking
for
and
also
with
the
replication
and
the
trivia
scanning
was
really
good
because
it
supports
suse
images.
D
So
those
were
a
few
of
the
things
that
we
that
were
on
the
top
of
our
list
when
we
were
selecting
while
we
were
looking
for
registry-
and
I
think
throughout
through
the
process
like
most
of
these
features,
have
been
working
pretty
smoothly
and
yeah.
We've
just
submitted
the
changes
for
the
minor
stuff
that
we
thought
could
be
improved.
Yeah.
The
journey
has
been
really
good
so
far,.
A
Great
yeah,
that's
that's
awesome,
that's
great
to
hear
so
is
this
registry?
A
D
So
we
have
tests,
so
we
have
our
own
kubernetes
platform
called
cars
which
we've
mainly
tested
on,
but
we've
also
tested
with
rancher
and
with
public
cloud,
that's
aps
and
eks.
So
we've
covered.
It
should
technically
work
on
any
kubernetes
deployment.
But
these
are
the
ones
that
we've
tested
on
and
it's
been
working,
fine.
B
D
Yeah
and
we've
rebranded
the
ui,
but
if
you're
interested,
we
can
share
the
screen
and
show
a
bit
of
that.
A
Yeah,
absolutely
if
you
want
to.
C
This
so
so
to
be
clear:
it's
just
theming
right,
so
we
did
not
modify
the
ui
in
any
way,
so
functionality
and
look,
and
so
on
is
the
same.
It's
just
like,
because
style
sheets
really
that
have
been
adjusted
to
the
too
much
corporate
feeling.
A
B
B
What
else
should
I
should?
I
share
sally
you've
been
working
more
on
this
than
I
have.
D
Yeah,
I
think
yeah
the
colors
and
some
of
the
buttons
that
have
been
changed,
not
not
much
to
the
ui
and
the
login.
E
C
So
as
of
today,
I
think
this
is
just
a
modification
that
we
carry
forward.
We
have
not
come
across
like
some
kind
of
feminine
concept
or
plug-in
concept
in
in
harbor.
I
think
currently,
the
way
we
toned
it
down.
It's
it's
not
a
huge
burden
to
carry
this,
but
potentially,
if
there's
like
others
that
have
to
need
to
like
customize
the
view
to
make
it
more
like
corporate
branded
for
like
their
deployment
or
so
then
yeah,
maybe
some
some
more
work
in
that
regard
would
be
useful.
A
That's
great
thanks
guys,
I
don't
know
if
anyone
else
had
any
questions.
A
A
That
all
right
yeah,
if
we
don't
have
any
other
questions
for
susie,
sorry
nope.
Sorry,
nothing!
Oh
that's
huge,
honest
signing.
I
Could
I
have
one
question
for
susie
fox
if
they
are
there,
this
is
daniel
from
aqua
security.
I
was
wondering
what
is
your
experience
with
the
interrogation
services?
I'm
always
curious.
What
is
the
feedback,
and
I
thought
it
would
be
nice
if
you
can
share
if
you
are
using
interrogation
services
or
maybe
what
is
the
biggest
struggle
if
you
could
share,
that
would
be
great
all
right,
so
you
have
3v
scanner
adapter
installed,
but
yeah.
C
So
currently,
for
me
answer
that
currently
we
are
just
following
the
the
harvard
projects
upstream
default
of
3d.
A
C
It
is
trivia,
has
a
scanner.
A
C
As
your
air
force
user
security
notifications,
so
it's
really
integrated
with
the
data
that
we
provide
for
any
distribution
updates
and
so
on
that
that
suse
is
providing
on
any
platform.
So
from
that
point
of
view
it's
nicely
integrated,
we
did.
I
think
so.
I
personally
like
submitted
like
one
or
two
really
minor
patches
to
like
improve
the
accuracy
of
the
detection,
but
so
far
we
have
not
done
further
for
a
work
on
this.
C
A
Yeah
by
the
way,
we're
making
trivia
the
the
only
default
scanner
that
comes
with
hardware
installations,
so
player
is
going
to
be
deprecated
on
the
2.2,
so
you
can
store
it
in
software.
That's
not
a
atmosphere
scanner,
but
we're
going
with
trivia
the
single
default
scanning.
So.
C
Yeah
we've
seen
that
in
the
announcement.
I
think
we
have
absolutely
no
concerns
about
it.
A
Okay,
great
yeah,
we
were
trying
to
we're
thinking
of
making
announcements
more
prominent.
Maybe
writing
a
blog
or
something
around
it,
because
for
existing
clear
users
you
will
have
to
essentially
reinstall
so,
but
it's
great
that
you're,
you
know
already
seen
it.
A
All
right
cool
thanks
thanks
to
everyone
from
susie,
for
for
doing
this.
If
you're
coming
on
here
and
talking
about
it,
it's
very
helpful,
very
encouraging.
A
H
F
H
Okay,
as
we.
H
As
we
know
in
the
harvard
2.2
we'll
introduce
the
performance
metric
features
in
hubble
and
actually
that
there
is
a
little
bit
of
change
from
the
first
design
and
in
the
currently
a
way
designed
to
we
decided
to
make
a
explorer
here
and
to
collect
the
information
from
the
hubble
database
and
to
show
the
hubble
related
information
and
every
component
we
will
expose
the
performance
metric
itself.
H
H
As
you
can
see
here
and
oh
yeah
actually
currently
to
enable
to
enable
this
feature
is
very
easy:
you
can
you
can
just
yeah,
we
have
a
new
config
item
in
the
hubble.
Then,
however
diamond
configure
file
and
there
is
a
metric
item
and
you
can
enable
it
and
set
the
port
and
the
path
and
the
harbor
will
expose
the
matrix
from
hold
and
the
metric
and
the
past.
E
H
Yeah,
yes
and
okay.
Let's
continue
and,
as
you
can
see
in
this
configuration,
we
open
a
18
and
one
point
port
and
matrix
as
our
path,
and
we
can,
through
this,
this
url
to
get
our
metrics
and
there
are
parameters
which
means
which
component
is
performance
metric,
for
example.
This
is
the
core
component
and
we
will
expose
all
the
metrics
we
collected
from
the
core
and
in
core.
H
We
enabled
the
goals,
memory,
states
and
the
goals
which
is
exposed
by
the
goal,
library,
promising
school
library-
and
we
also
add
some
performance
metrics
in
hubble,
for
example,
the
request
number
of
each
url
and
the
duration
of
each
request.
For
example,
the
hardware
called
http
requisition
second,
which
is
a
summary
of
all
the
performance
data
of
the
request.
H
Time
and
if
you
wanna
say
that
as
a
component,
you,
for
example,
the
registry
you
can
just
change
the
parameter
this
matrix
is-
is
exposed
by
the
registry,
which
is
the
docker
distributed.
H
Industry,
distribution
behind
the
harbor-
and
we
also,
if
you
remove
this
component
of
this,
this
matrix,
is
exposed
by
the
hardware
exporter
and
you,
as
you
can
see
in
the
hardware
exporter,
we
expose
some
a
lot
of
hubble
projects
related
metrics,
for
example,
the
image
polled
and
the
effects
tell
the
number
of
artifacts,
and
we
and
something
like
this.
H
H
Oh,
you
can
say
that
there
is
the
health
state
which
is
collected
from
the
power
core,
and
this
is
quota
usage,
which
is
called
this.
Data
is
collected
from
the
harvard
database
and
something
else
like
image.
Protocol
revocable
effect
account
and
the
performance
metrics,
for
example.
H
This
is
a
the
garbage
in
our
memory
in
our
service
memory,
and
you
can
say,
the
label
which
is
identified
by
the
label
of
each
each
line
means
a
component
memory
cabbage
of
gold,
and
there
is
also
your
uri
urls
request
data,
for
example.
This
is
the
request
rates
of
each
path
of
harbor
core,
or
something
like
this.
C
So,
for
me
this
looks
really
great
thanks
for
sharing
one
question
that
I'm
that
I'm
curious
about.
Do
you
track
things
like
a
number
of
failed
calls,
or
so,
like
sorry,
percentage
of
like
failed
api
calls
or
failed
replications
yeah.
H
H
E
The
idea
is,
you
can
write
your
from
ql
to
you
know,
query
monitors
in
the
failures.
C
A
Yeah
this
I
mean
this
is
the
first
pass
at
some
of
the
metrics
that
I
think
people
will
be
interested
in
and
we'll
refine
them
over
time.
F
And
there
was
another
project
that
I've
saw
on
on
github,
which
is
the
exporter
for
harbor,
but
harbor
matrix
for
data
like
counter
repositories
and
everything
and
the
person
who
was
responsible
for
that,
for
that
was
asking
if
it
could
be
kind
of
emerged
into
harbor
as
well.
So
I'm
not
sure
if
you
were
aware
about
this,
and
you
opened
up
a
much
request
for
that
as
well.
So
I'm
not
sure
if
it's
kind
of
a
there
is
an
agenda
for
that
as
well.
H
Oh
actually
yeah,
we
we
know
there
is
some
projects
on
github
to
expose
the
hubble
metrics
and
well.
We
I
think
we
didn't
have
any
contacts
with
the
authors
and
so
there's
no
plan
to
merge
better.
Maybe
we
can
exposed
some
metrics
like
maximum
value.
I
I
think
that
that
kind
of
explode
only
expose
the
metrics
of
which
can
get
from
the
hubble
api,
and
we,
you
know
some
performance
metrics.
We
need
to
using
instrumenting
pointer
to
in
our
code
to
collect
the
performance
metrics.
H
So
I
think
in
our
version
we
can
expose
more
match
more
performance,
metrics.
E
A
A
On
by
the
someone
over
systick
who
we
have
a
contact
with,
I
think
we
I
looked
at
a
little
bit,
but
it's
gonna
be
added
in
in
stages.
So
if
it
makes
sense,
then
I'm
sure
we'll
be
able
to
add
it.
E
C
C
E
A
Thanks
duncan,
so
this
is
looking
really
good,
something
that
we've
been
trying
to
do
for
the
longest
time
and
I'm
glad
that
we
finally
got
to
it
so
yeah.
It's
it's
something.
It's
definitely
needed
for
for
any
modern
teams
that
are
that
are
running
something
like
the
registry.
A
F
Yeah,
thank
you
very
much.
Thank
you
for
being
part
of
this
community.
Let
me
just
give
you
just
a
quick
overview.
What
we're
doing
so.
We
we're
doing
a
container
registry
service
for
actually
four
or
five
years
already,
and
we
started
as
a
it's
more
side
project.
We
started
with
a
container
registry.
It
was
based
on
portas
and
we've
been,
you
know,
making
contribution
to
portrait
the
portals.
F
F
It's
basically
a
one-to-one
hardware,
so
there's
no
customizations
made
a
bit
of
customization
is
there,
but
it's
not
something
that
yeah,
maybe
one
feature
I
would
like
to
open
up
for
as
a
pr
which
is
yeah,
it's
a
tiny
one,
but
normally
it's
it's
a
manila
harbor,
which
we
kind
of
yeah
hardened
a
bit
for
our
infrastructure,
and
we
are
targeting
you
know,
companies
which
are
also
on
aws
and
cloud
providers.
So
mostly
we
are
working
with
cloud
providers
who
would
like
to
bring
this
in
the
in
their
cloud
infrastructure,
hyper-based
power
based
solutions.
F
So
we
help
them.
You
know
kind
of
do
harbor
on
scale.
F
So
this
is
our
our
focus
and
we
would
like
actually
to
dedicate
more
time
and
more
resources
into
contributing
to
hover
in
in
the
near
future.
So
we
would
like
to
have
a
maybe
more
more
people
working
on
harbor
and
on
contributing
back,
especially
what
we
were
thinking
on
long
term
is
having
a
true
multi-tenant
harbor,
where
right
now
there
is
no
really
multi-talent
harbor,
but
I
mean
it's
kind
of
a
multi-ten,
but
it's
not
true
true
multitask.
F
So
maybe
this
is
something
we
might
be
interested
in
contributing
and
our
our
main
goal
is
really
do
everything
that
is
well
useful,
contribute
back
to
harbor,
because
we
think
it's
I
mean
we
owe
it
and
also
we.
We
think
it
will
benefit
us
more
than
keeping
it
for
ourselves.
F
So
if
we,
if
we
contribute
back-
and
so
we
would
like
to
as
a
default
strategy,
we
will
contribute
everything
back
to
to
to
the
community
unless
it's
not
needed
or
it's
something
that
you
know
it's
just
something
specific
to
us
only
and
it
makes
no
sense
but
normal.
So
we
we
plan
to
contribute
everything
back
to
the
community
and
we
already
identified
things
where
we
would
like
to
improve-
or
you
know,
make
things
better,
especially
for
day
one,
because
we,
you
know
we
try
to
operate
harbor
at
scale.
F
F
A
E
I
think
it's
particularly
important
you
mentioned
that
multi-tenancy.
Actually,
we
have
a
lot
of
internal
debate
regarding
what
should
we
do
in
regards
to?
You
know
better
supporting
multi-tenancy.
I
totally
agree
that
harbor
is
not
really
a
multi-tenant
software,
so
your
your
import
in
this
part
will
be
extremely
valuable
to
us.
F
Yes,
yes,
I
mean
we,
we
have
it
an
agenda
because,
right
now
we
we
we
do
actually
deployments
of
harbor.
You
know
we
manage
it
automatically,
but
it's
a
I
mean
we're
at
a
few
hundreds
of
hybrid
installations
and
we
plan
to
reach
a
thousand.
So
at
that
point
we
need
to
have
a
kind
of
a
do,
a
contribution
regarding
a
true
multi-tenancy
setup,
correct.
E
F
A
F
Well,
because,
actually
I
mean
it's
it's,
it
has
something
to
do
with
the
the
setup
that
we
have
and
also
that
the
experience
with
in
the
past
I
mean
we
had
in
the
past
with
reporters
the
experience
that
we
rushed
a
bit
too
much,
and
so
we
developed
a
lot
of
things
for
on
top
of
portland
that
we
couldn't
backboard
anymore,
because
they
were,
you
know
so
with
there's
also
too
much
drift,
and
we
wanted
to
avoid
this.
F
We
wanted
really
to
keep
it
clean
as
possible,
so
we
wanted
to
now
to
have
a
close
as
harbor
as
possible,
so
that
if
harbor
is
releasing
a
new
release,
we
a
few
days
later,
we
can
follow
you
know
and
for
for
the
clients
that
we
were
targeting
it's
important
to
have
their
own
identity
access
management
and
also
their
own
domain,
and
so
this
is
the
only
way
how
we
could
achieve
this
with
with
harbor.
F
At
the
moment-
and
I
mean
we
have
knowledge
of
kubernetes,
so
for
us
it's
I
mean
it's
a
challenge,
but
it's
it's
doable.
So
we
can
handle
those.
Those
instance
management
at
scale.
A
Right
I
mean
I,
I
think
yeah
I
would
agree
it's
it's
not
multi-tenant
in
a
larger
scheme
of
things,
but
I
think
it's
still
built
to
handle
multiple
clients
or
customers
or
teams.
So
yeah,
if
you
want,
we
can
deep
dive
this
and
you
know
maybe
talk
about
like
recommended
practices,
things
of
that
sort
to
see.
If
do
you
really
need
hundreds
of
hybrid
installations
to
manage
your
your
clients
or
customers.
F
Yeah
it's
with
the
identity
management.
So
if,
if
we
want
to
have
like
this,
you
know
custom
identity,
management
right
now,
yeah,
it
is
possible,
so
we've
been
investigating
this
topic
already,
how
to
make
it
multi-tenant
and
what
needs
to
be
changed
with
components.
So
we
are
looking
already
into
this
topic
and
we
won't
move
away,
but
you
know
right
now:
we
want
to
move
forward
already
and
yeah.
So
this
is
actually
the
our
current
strategy
and
I
would
be
more
than
glad
or
happy
if
we
don't
have
to
manage
them
all.
H
A
There
is
a
work
stream
on
deploying
multiple
harbors,
like
the
at
the
scale
you're
talking
about
hundreds,
thousands,
tens
of
thousands,
it's
mostly
for
registries
at
the
edge,
so
we're
working
on
a
light
installation
of
harbor.
It's
called
like
harbor
light.
It's.
A
A
strip,
a
registry,
that's
stripped
down
to
some
essential
features
for
something
at
the
edge.
You
know
where
they're
not
they're,
mostly
on
the
receiving
end,
getting
images,
probably
not
that
many
requirements
in
terms
of
image
scanning
and
signing
they're.
Mostly,
you
know
images
they're,
consuming
images
being
replicated
from
a
central
spoke.
So
that's
something
we
can
talk
about
as
well.
A
All
right,
yeah,
I
see
daniel's
comment
about
legacy.
Robot
accounts.
I
don't
think
we
have
enough
time
today
because
it's
a
pretty
long
discussion,
essentially
we're
going
to
be
rolling
out
harvard
system
level,
robot
accounts
in
2.2.
A
So
I
don't
have
anything
else
for
today.
Does
anyone
have
any
comments
or
questions
anything
else?
You
want
to
ask
ask
what
dean
asked
that
the
folks
said,
susie.
A
All
right:
well,
thanks
everyone
thanks
for
the
discussion,
it's
really.