►
From YouTube: Harbor Office Hours - Oct 28, 2020
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
A
Are
live
so
yeah
welcome
again
to
the
harbor
office
hours
here.
Joby
did
you
have
any
questions.
C
C
So
the
thing
is
we
want
to
integrate
hardware
with
the
gitlab
ci
using
tokens,
and
I've
been
going
around,
checking
the
github
repository
and
the
issues
and
all-
and
there
have
been
quite
a
few
issues
regarding
this,
but
there's
not
a
solution
like
a
complete
solution
from
harbor
side.
So
that's
what
I
wanted
to
discuss
here.
C
B
Yeah,
I
think
it's
good.
You
mentioned
that
I
actually
know
a
customer
from
us
which
was
doing
or
trying
something
similar
a
few
months
ago.
I
think
we
just
have
to
re-prioritize
that
with
the
horrible
team.
I
guess
I
mean
that's,
not
something
I
I
can
answer
right.
If
it's
not
working,
it's
not
working,
and
I
actually
didn't
test
it
myself
to
see
the
issue.
So
there
are
tickets,
open
right,
you
said,
or
is
that
not
the
case?
C
Yeah
there
are
two
like
issues
open.
If
you
want,
I
can
add
those
issues
on
the
chat
and.
B
C
B
Yeah,
I
can
actually
show
you
how
I
just
set
up
harbor.
I
think
five
minutes
before
this
call
how
I
did
it.
I
mean
there
are
many
possibilities
to
do
that.
Our
own
testing,
I
think,
is
using
the
offline
installer,
because
I
mean
it's
very
easy
to
script.
B
You
just
have
a
linux
vm
or
a
host
or
whatever,
and
you
just
have
a
few,
let's
say
steps
to
go
through.
I
can
show
you
that
in
a
second,
if
you
have
kubernetes
running
somewhere,
of
course,
you
can
use
the
helm
shorts,
I
mean
there
are
different
ones,
the
I
would
say
the
official
ones
and
also
the
ones
from
vietnamese
which
normally
have
more
stuff
added,
and
this
is
also
really
fully
tested.
B
The
third
option,
this
one.
Actually,
I
cannot
show
yet
because
it's
really
new
and
I
didn't
play
with
it-
myself-
is
using
the
harbor
operator.
I
think
that's
probably
the
future
of
of
also
the
insulation
on
kubernetes,
because
it
can
have
postgres,
high
availability
and
all
these
these
components.
B
In
the
back
end,
I
mean,
if
you
have
to
set
up
that
on
your
own.
It's
it's
a
lot
of
work
and
the
harbor
helmet
shots.
Don't
do
that,
but
the
operator
framework
is,
I
would
say,
the
future
for
the
installation,
and
I
mean
there
is
other
other
ways
there
is
a
tile
in
in
one
of
our
products.
I
mean
this,
but
this
is
not
really
the
point
here,
which
is
using
bosch
to
set
up
the
vms
and
install
harbor.
B
So
basically,
I'm
testing
all
the
free
setups
right
now
and
the
operator
is
on
my
list.
But,
as
I
said,
I
can
show
you
how
I
do
it.
It's
it's
really
simple.
Then
the
offline
installer
is
is
really
simple
to
use,
but
of
course
you
have
to
set
up
the
certificate
service
right.
First
right
I
mean
that's
the
major
step.
If
you
don't
use
certificates
and
all
this
stuff
all
day
long,
I
mean
each
time.
I
have
to
really
look
up.
B
The
open
ssl
commands
to
do
it
on
my
own
or
use
other
frameworks,
but
anyway,
just
put
them
in
the
chat
that
would
be
nice
to
see
the
tickets,
because
that
was
actually
I
mean.
I
don't
know
in
one
of
the
announcements
we
had
that
this
should
work,
so
I'm
actually
surprised
that
it
doesn't,
but
that
would
help
me
if
you
send
me
the
link
to
the
to
the
issues.
B
So
yeah
my
installation,
actually
I
just
had
I
mean
you
see
it's,
I
didn't
write
a
script
yet,
but
this
is
actually
all
the
steps
are
just
executed.
I
can't
do
it
again,
but
I
mean
probably
no
sense
of
doing
that.
I'm
just
creating
my
own
certificate
authority,
creating
a
signing
request
with
openness,
ssl
and
then
signing
it
with
my
own
one,
and
this
is
just
certificate
stuff
right
I
mean.
There's
you
probably
have
your
own
way
of
doing
that.
B
What
I
normally
do
is
here
to
add
alternate
names,
because
I
see
many
pro
also
products
from
companies
which
are
not
doing
that
right,
so
I
can
actually
access
it
through
the
full
dns
name,
the
short
name
and
also
the
ip
address.
I
mean.
That's,
that's
really
important
and
basically
I
copy
all
the
stuff
to
the
slash
data
directory,
because
that's
where
harvard
will
actually
save
all
the
the
other,
the
data
and
you
also
have
to
copy
you
see
it
here
in
that
step,
the
certificate
you
created
above
to
that
download
directory.
B
B
Thank
you
that
helps
indeed,
so
let's
go
for
the
steps
again
so
again
here
right
I
mean
I
create
my
ssl
directory,
I'm
creating
a
certificate
authority
and
that's
something
actually
on
the
ubuntu
system.
I
I
was
using,
which
sure
was.
I
didn't
know
that
you
have
to
create
this
random
file,
just
touch
it.
Otherwise,
the
open
ssl
will
give.
You
error
messages,
it's
weird,
but
that's
how
it
is
so
I
create
my
certificate
request,
signing
request.
B
I
sign
it
and,
as
I
said,
this
is
a
step
I
was
just
explaining
alternate
names
that
I
really
use
the
dns
names
and
the
ip
address,
so
it's
really
fully
accessible
the
way
I
want
it
and
then
basically,
I
copy
everything
to
this
data
assert
directory
that
that's
important
and
also
this
certificate
to
the
download
page.
I
will
show
that
in
a
second
because
then
you
can
go
to
the
ui
and
and
download
it
or
do
a
curl
command
or
whatever
so
and
yeah.
B
Here
you
see
where
I
copy
all
this
stuff
right
I
mean
it's
all
explained
in
the
documentation,
which
is
very
good
from
harbor,
but
the
certificate
stuff.
Honestly,
personally,
I'm
always
struggling
with
that.
Sometimes
it
works.
Sometimes
not
if
you
look
at
documentation
from
different
parts,
it's
always
a
little
bit
different
and
sometimes
I'm
actually
struggling
myself
to
get
it
running,
but
anyway
I
mean
I
can
put
that
somewhere
on
github.
B
If
you
want
to
have
that
or
maybe
in
the
in
the
meeting
notes,
then
I
get
the
offline
installer
I
mean
just
the
latest
one,
I'm
really
always
on
the
latest.
You
have
to
know
here
and
that's
probably
important.
You
know
we
have
releases
right
this.
These
are
tested
and
the
update
mechanism
is
also
tested.
B
So
you
know,
when
you
go
from
hardware
to
zero
to
one
that
actually
the
upgrade
will
work,
because
it
could
be
that
in
the
postgres
database
you
are
changing
some
properties
or
adding
things
whatever.
So
we
are,
we
are
changing
the
schema
of
the
database
and
this
is
only
really
tested
from
ga
to
ga
release
and
between
them.
With
this,
let's
say
latest
build.
This
is
not
done
and
it
could
be
that
it
will
fail
and
then
it
will
fail
forever.
B
Right
I
mean,
then
you
have
to
go
and
actually
delete
this
data
directory
or
the
postgres
data,
so
it
will
be
recreated.
That's
something
I
run
into
myself
and
I
really
didn't
understand
that.
So
that's
all
you
unpack
you
unpack
it.
Then.
Basically,
you
go
to
the
harbor
yama
configuration
and,
of
course,
that's
a
whole
thing
in
itself.
Right
I
mean
there
are
many
many
properties
you
can
set
and
the
the
helm
chart
does
the
same
right.
B
I
mean
you
just
specify
it
differently
what
I
normally
add,
as
you
see
here,
the
certificate
pass,
because
you
need
that
maybe
a
password
and
yeah
some
other
stuff
log
rotation.
B
If
you,
if
you
the
components
here,
you
see
it's
a
3v
clear
for
the
scanning
engine,
sub
engines,
job
services
core,
but
as
I
said,
I
just
use
that
as
my
default
and
then
you
basically
do
an
install
with
trivia
with
short
museum
with
clary
notary.
So
you
have
everything
just
wait,
but
you
need
docker
compose
to
to
be
installed
for
that.
So
that's
actually
what
did
not
work
just
five
minutes
ago,
because
when
I
do
an
opt-get
install
docker
compose
it
installed
the
version
here
on
ubuntu
1804,
probably
an
ubuntu
2004.
B
It's
it's
not
the
case
anymore,
which
was
outdated,
so
I
actually
had
to
install,
I
think,
yeah,
the
latest
local
composer
installed
it
manually
here.
B
B
Yeah
docker
proxy:
that's
the
process
running
and
that's
it
just
copy
it.
I
get
the
cert,
I
mean
that's
what
I
said
you
you
actually
need
to
copy
to
this
directory.
Otherwise
you
cannot
curl
it
and
you
see
just
an
api
call.
I
mean
we
can
go
to
the
ui
to
do
the
same
thing.
Whether
I
just
use
the
api.
B
And
then
I
logged
in
into
this
instance
I
just
created,
and
you
see
that
that's
it
I
mean
that's.
What's
the
manual
installation,
it's
sometimes
not
bad
to
start
with
the
manual
one
right,
because
you
really
see
what's
going
on,
if
you
do
the
the
helm
installation,
it's
also
very
simple,
you
just
have
to
know
a
few
things
that
yeah.
I
think
this
is
the
command
which
actually.
B
Let's
see
have
it
once
more:
where
is.
B
B
Yeah,
I
think
here
that's
a
bit
nami
harbor
helm
chart,
so
I
mean
you
have
to
set
a
few
things:
the
storage
class.
Maybe
it's
good
if
you
need
special
storage
for
that,
the
service
common
name
and
also
the
external
url.
I
mean
this
is
important.
Otherwise
you
cannot
access
it.
So
you
basically
set
the
properties
you
do
in
this
yaml
file
for
the
help
configuration
you
do
it
on
on
the
command
line
or
you
create
a
values
file
and
that's
it
right
and
then
the
certificate
all
this
stuff
is
done
internally.
B
You
don't
have
to
handle
that
the
the
help
chart
is
certainly
easier
to
install
than
the
one.
So
going
back
to
to
what
you
said,
if
you
have
it
installed,
I
mean
you
may
might
do
a
replication
to
your
other
registry
and
just
try
to
get
the
stuff
over,
but
you
can
also
just
upload
the
the
container
images
to
your
hardware
registry
and
configure
it
right.
I
mean,
as
you
have
said,
the
the
user
configuration
did
not
work
with
the
tokens
from
github,
but
that's
something
we
can
actually
let's
say,
investigate
and
see.
B
What
is
it
called?
Yeah
open
api
and
you
can
execute
it
here.
You
see
all
the
stuff,
I
mean
applications
scheduling
also,
you
can
add
robo
accounts.
Actually,
if
you
want
to
do
some
automation,
I
mean
there's
a
lot
of
yeah
here,
the
robot
stuff.
So
there's
a
lot
of
things
going
on
there.
I
don't
know
the
whole
api
to
be
honest,
because
it's
pretty
it's
pretty
overwhelming
scanners
retention.
B
How
much
you
want
to
actually
keep
because
otherwise
you
will
blow
up
your
registry
and,
as
you
said,
I
mean
if
you
really
try
to
to
go
hard
against
the
registry.
I
would
really
be
interested
also
to
to
see
how
how
much
low
they
can
actually
handle.
B
I
mean
we
do,
of
course,
some
testing,
but
I
think
some
real
use,
I
mean
some
real
use.
Cases
would
be
interesting
to
hear
about
if
you
move
over
so
yeah.
That
was
just
a
quick
introduction
to
how
you
can
install
it
by
hand.
As
I
said
it
just
took
a
few
minutes
to
to
do
that.
With
my
my
commands
I
had
here
and
then
you
can
play
with.
E
E
Sure
so
yeah
we
are
using
harbor.
Now
we
love
it
for
both
images
and
helm
charts.
I
was
curious.
I
haven't
seen
it
in
the
in
the
github
issue
list.
I
was
curious
if
it
ever
came
up
in
the
past
about
possibly
supporting
generic
files.
So
so
we
have
a
requirement
to,
in
addition
to
helm,
charts
and
and
images
also
store
some
some
files
like
scripts
or
little
helper
tools
and
in
a
perfect
world
we
could
have
one
one
registry
to
rule
them
all.
E
But
if,
if
that's
not
in
the
plan
ever
then
we'll
just
we'll
just
stand
up,
I
guess
a
simple
registry.
Our
simple
file
share
registry,
alongside
harper.
E
E
E
B
Yeah
yeah,
I'm
not
aware
of
any
plans
right
the
question.
Of
course
the
team
will
ask
us
why
you
want
to
do
that.
You
just
put
everything
in
a
git,
gitlab,
github,
repo
and
whatever,
and
do
the
the
versioning
and
all
this
stuff,
because
if
he
would
go
down
that
path
with
harbor
I
mean
we
are
opening
a
huge
valley
so
to
say-
and
I'm
not
sure
that's
really
the
goal
of
harbor.
But
anyway
I
took
some
node
and
we'll
see
what
our
development
teams
tell
us
about.
That.
E
Yeah
and
just
to
flesh
out
our
use
case
a
little
more,
so
we
are
actually
using
harbor
to
deliver
cloud
native
artifacts
to
our
customers,
so
the
so
so
they
are
so
we
have
harbor
integrated
with
octa
for
user
management
and
so
for
us,
we're
going
to
need
a
file
registry
also
integrated
with
octa
to
let
our
the
specific
customers
that
are
entitled
download
these
other
tools.
I
was
mentioning
like
simple
scripts
or
helper
tools
or
whatever.
E
So,
yes,
we
could
use
sort
of
an
anonymous,
read
github
repo,
but
we
have
some
artifacts
that
are
not
images,
not
helm,
charts,
but
clients
would
still
need
to
be
able
to
pull
so
that
that's
really
our
use
case.
It's
you
know
it's
pretty
simple
for
us
to
stand
up
a
generic
file
share
repo
do
with
nginx
or
whatever,
but
yeah.
It's
mainly
just
curious.
If
that
was
anywhere
on
your
road
map
sounds
like
it's
not
just
fine.
A
I
would,
I
would
definitely
open
an
issue
and
explain
your
use
case,
because
that
way
we
can,
we
can
track
it
and
see
if
we
can
do
something
there.
E
Great
we'll
do
the
other
question
we
had
we're
just
about
to
upgrade
to
2.1
we're
looking
at
it
right
now,
we're
on
2.0
now
just
curious.
If
there's
any
gotchas
or
anything,
we
need
to
worry
about,
or
is
it
supposed
to
be
pretty.
D
B
Just
going
one
second
back
the
I
had
other
customers
from
us
actually
asking
a
few
months
ago.
Almost
the
same
use
case
you
were
just
describing,
I
just
forgot
it,
so
I
mean
it's
something:
you're,
not
the
first
one
actually
trying
or
wanting
to
do
that
with
the
files,
because
they're
also
distributing
actually
files
to
their
own
customers
and
would
like
to
do
it.
B
B
Anyway,
it's
a
discussion
we
can
have
when
you
open
the
issue
with
the
development
team
to
see
and
and
regarding
updates,
I
said
that
before
I
mean
the
when
I
showed
the
offline
installer,
the
the
updates
between
intermediate
versions
right
between
ga
version
is
not
really
tested,
but
between
j
versions,
the
heavy
tests
I
mean
they
work,
I'm
not
aware
of
any,
because
I'm
updating,
I
think
I
just
updated
one
two
hours
ago
and
I
didn't
have
any
issues,
but
you
never
know.
B
B
Yeah,
okay,
and
did
you
use
the
the
the
helm
wrapper,
the
official
one
or
the
beaten
army,
one
which
is
used
a
lot
also
by
community
folks.
D
I
I
don't,
but
shashank
is
also
on.
He
should
be
able
to
answer
yeah.
Even
I
don't
remember,
and
that.
B
That
people
are
hitting
no,
I
mean
technically.
The
only
thing
which
can
go
wrong
is
when
we
update
the
schema
of
the
postgres
database,
or
things
like
that
and
but,
as
I
said,
this
is
really
tested
from
ga
to
ga
release.
So
I'm
not
really
expecting
anything
there.
If,
if
you
do
what
I
did
do
basically
a
weekly
update,
you
can
you
can
break
things
right,
because
this
is
not
tested
and
the
schema
upgrade
can
fail
and
then
your
harboring
selection
is
messed
up
and
you
don't
know
why?
B
E
Yeah
and
we're
looking
forward
to
a
few
features
in
there,
like
the
one
is
because
we
host
harbor
for
customer
download,
we
always
kind
of
struggled
in
the
past,
because
the
the
user
got
to
pick
their
local
harbor
user
name,
which
we
didn't
like.
We
always
wanted
it
to
be
their
email
address.
So
it
looks
like
with
2.1.
E
We
can
now
lock
that
down
and
just
always
make
it
the
username
and
opera
or
the
email
in
octa,
which
is
great,
yeah,
perfect,
oh,
and
by
the
way
I
didn't
say
off
the
top
awesome
idea
hosting
these
office
hours
by
the
way
we're
we're
going
to
steal
this
idea
for
our
own
customers,
because
I
think
it's
a
great
way
to
let
customers
drop
in
on
their
on
their
own
time
and
and
also
time
boxes
it
for
you
guys
you
guys
are
just
on
for
two
hours
and
you
get
back
to
work.
B
Yeah
and
as
I
said
you,
you
can
really
ask
for
stuff
for
next
time,
maybe
I
mean
so
we
can
prepare
right.
I
mean
this
time
was
for
me
a
little
bit
hard
because
I
didn't
know
what
I
had
to
expect
and
what
kind
of
questions
and
of
course
you
don't
know
anything.
No,
we
don't
know
everything.
So
sorry,
so
sometimes
it's
good.
If
you
can
prepare
for
something.
B
If
you
want
to
see,
I
mean
the
the
one
we
just
discussed
with
with
the
token
that's
actually
a
good
one,
you
can
try
to
figure
out
why
it's
not
working
and
the
development
team
normally
reacts
pretty
fast.
If
there's
a
really
use
case
behind
it.
Yep.
E
Okay,
I
actually
did
have
another
question.
I
know
there's
a
an
issue
for
this
one.
Sorry,
maybe
I
have
it
here
bookmarked,
oh,
maybe
I
don't.
There
was
an
issue
I
saw
it
got
moved
to
2.2
plan
and
it
was
about
more
granular
access
control
beyond
just
the
project
level.
I
was
curious
if
that
work
had
started.
B
B
Yeah,
what
what
kind
of
more
granularity
are
you
looking
at
or
wishing
for.
E
Yeah,
so
because
we're
because
we
are
using
harbor
to
support
end
users,
our
end
users
are
entitled
to
different
parts
of
our
portfolio,
and
so
in
a
perfect
world,
we
would
set
up
our
harbor
registry
to
reflect
that,
and
so
users
would
get
read
access
and
actually
we
do
read
access
through
through
octa
groups,
but
they
would
get
read
access
to
just
the
stuff
they're
allowed
to
download.
So
today
the
only
way
we
can
implement
that
is
at
the
project
level.
E
So
we
have
to
have
a
separate
project
for
everything:
that's
separately
titable.
If,
if
it
was
more
granular
than
that,
then
maybe
we
could
structure
our
projects
by
key
function
or
by
business
group
business
area
and
we
could
actually
control
who
can
download
what
within
a
project.
So
we
don't
have
really
any
expectations
or
requirements
on
how
that
might
work.
I
was
more
curious
just
if
you,
if
you
had
an
idea
of
how
it
might
end
up
looking
or
if
it
hasn't
been
started.
Yet
that's
fine.
B
Yeah,
I
don't
know,
but
what
you
say
is
basically
request
number
one.
I
would
ask
myself
right:
I
mean
you
have
a
hierarchy,
so
you
have
customers
and
you
have
projects
and
business
lines
whatever
inside
that
project
right
and
you
can
give
permissions
to
that.
I
I
know
this
discussion
from
stack
times
many
many
years
ago.
I
mean
it
was
the
same
thing:
how
you
structure
all
this
stuff.
B
I
also
made
a
note
because
I
think
that
that's
that's
a
very
valid
requirement.
You
you
just
mentioned,
but
I
don't
know
actually
about
the
work
going
on
there.
B
B
They
really
wanted
to
have
quarters
and
limiting
and
whatever,
and
that
actually
is
now
in
there,
but
I
still
think
there
could
be
some
more
things
like
network
limits
or
whatever
I
mean,
did
you
run
into
any
issues
with
that
that
really
customers
are
are
doing
too
much
there?
I
mean
consuming
too
much
space
uploading
too
many
images.
I
mean
the
ap
api
limit
so
or
are
you
handling
that
differently?.
E
Yeah,
the
only
the
only
limits
we
would
hit
would
be
on
the
pull
side,
because
we
don't
allow
customers
to
push
anything
into
harbor.
That's
all
us
hosting
content
and
from
a
push
point
of
view
we
haven't.
I
don't
think
we've
seen
any
performance
issues
or
anything
we're
hosting,
really
large
images
like
we.
We
have
some
ridiculously
large
images.
Like
one
image,
I
saw
had
two
different
layers
that
are
each
five
gigs,
so
the
one
image
on
its
own
was
like
12
or
13
gigs,
just
ridiculous.
E
So
we
have
work
on
our
own
to
kind
of
clean
up
our
our
own
stuff,
but
from
a
poll
point
of
view,
yeah,
it's
basically
just
network.
We
have
tests
running
every
day
pulling
stuff.
I
don't
think
we've
seen
any
issues
yet,
but
at
the
same
time
we
haven't
scaled
up
to
like
thousands
of
users.
Yet
we
have
we
have
a
couple
and
we're
we're
just
getting
going
so.
Okay,
maybe
we
need
to
start
running
some
scalability
tests.
B
Yeah,
because
the
request
I
had
three
years
ago
was
for
four
thousand
users
and
I
got
a
little
bit
nervous
right.
I
mean
four
thousand
users
going
to
one
registry.
That's
not
gonna
work
right
and
you
need
rate
limiting
of.
Of
course
you
you
can
do
things,
I
mean
with
overlays
and
networking
I
mean
you
can
do
stuff,
but
I
think
it
should
be
part
of
harbor.
Also
at
some
point
so
yeah.
It
was
just
a
question.
E
Yeah
and
another
another
aspect
of
that
which
probably
doesn't
matter
to
you
guys
too
much,
but
to
us
it
does
because
we're
we're
hosting
in
gcp,
and
so
we
pay
data
transfer
costs
for
every
byte
that
gets
transferred
out
of
gcp
into
or
out
of
that
network
into
another
network
and
so
yeah,
and
we
might
have
to
find
ways
to
limit
customer
polls
as
well
like
if
they,
if
a
customer
creates
their
own
ci
cd
pipeline
that
pulls
a
thousand
times
a
day.
E
That's
pro
that
costs
us
money
every
time
they
do
it
so
yeah.
We're
gonna
have
to
look
at
that
as
well.
So,
actually
on
that
point,
has
anyone
talked
about
harbor,
like
in
edge
environments
or
in
sort
of
like
cloud
delivery,
network
type
things
or
pushing
images
out
to
geographically
different
regions?
Things
like
that.
B
Yeah
I
mean
this
is
not
community-minded,
but
I
know
vmware
has
published
a
white
paper
about
edge
deployments
and
actually
the
harbor
use
case
is
actually
documented
that
the
the
problem
is,
of
course,
if
you
have
many
many
edge
edges
right,
I
mean
many
locations
and
you
have
maybe
one
central
harbor
registry.
I
mean
you
will
get
a
lot
of
traffic,
that's
generally
a
problem
how
to
to
distribute
things,
but
regarding
distribution
of
images
there
are.
There
are
many
harbor
projects
coming
along
how
to
do
that
in
a
very
efficient
way.
B
I
mean
using
peer-to-peer
technologies
and
I
mean
not
one-to-one
push
right,
so
you
just
broadcast
all
this
stuff,
somehow
I'm
not
sure
what
you're
what
you're
looking
for.
I
mean
the
rates,
the
rates
and
the
amount
of
gigabytes
to
transfer
to
many
locations
that
that
could
be
overwhelming
right.
Then
you
have
to
build.
B
E
And
actually
some
of
our
users
like
we
have
the
you
know
the
big
enterprise
customers
or
our
users,
like
banks,
and
all
of
that
a
lot
of
them
are
very
hesitant
to
allow
their
own
employees
to
pull
from
public
registries
or
even
registries
on
the
internet,
so
they
host
their
own
internal.
So
our
hope
is
that,
yes,
we'll
say:
here's
all
our
images
for
the
product
that
you've
just
purchased.
Mr
mrs
customer
and
your
first
operation
is
to
replicate
or
clone
the
images
to
your
local
registry
and
then
always
pull
from
there.
E
B
I
mean
you
know
right
now:
there
are
many
api
limits
which
are
becoming
active
for
docker
and
github.
Yes,
I
can
tell
you,
I
run
into
this
issue
daily
and
we're
also
trying
to
figuring
out
how
to
handle
that
with
local
harbor
instances
and
things
like
that
for
stuff.
We
use
all
the
time
because
all
my
little
tests
I
do,
I
basically
run
into
these
limits
all
the
time
right
now.
Yeah.
E
B
Forward
to
that
too,
that's
the
most
important
feature
ever
I
mean,
because
I
was
asking
for
that
personally
for
years.
Actually,
that's
a
big
thing,
because
then
you
can
also,
I
mean,
scan
all
the
images,
even
if
they
come
from
the
internet.
I
mean
I'm
always
surprised
how
how
red
all
this
stuff
is,
which
I
I
get
in
my
testing
environments.
E
D
Yeah,
unfortunately,
for
the
for
you
guys,
I
do
have
another
question.
There's
a
another
issue
we've
hit,
which
I
can
paste
the
github
issue
in,
so
that
you
guys
can
take
a
peek.
A
Just
one
one
quick,
so
jeff,
you
asked
about
scripts
so
right
now,
I'm
chatting
with
the
pm,
unfortunately
he's
in
another
meeting.
Otherwise
he
would
be
here
because
you
all
have
fantastic
questions
so
scripts
and
and
standalone
files,
not
right
now
focused
on
the
oci
wrapper
around
it.
Yep
yep
makes
sense.
D
No
no
worries,
so
we
initially
started
using
redis
h
a
and
ran
into
a
log
and
panic
and
then
backed
up
to
using
a
single
ratta.
So
the
basic
question
is:
have
you
guys
looked
at
this
issue,
any
ideas
on
it?
We
would
like
to
use
the
aha
version,
but
I
mean
we
clearly
have
a
workaround,
but
just
wanted
to
know
if
there
was
an
update.
B
Interesting,
I
wonder
what
the
the
operator
you
know
the
hubble
operator,
which
is
now
yeah,
let's
say,
being
improved
a
lot
and
how
they
handle
that,
because
I
mean
it
looks
like
since
you
set
up
everything
in
an
h,
a
fashion
we'll
run
into
this
issue.
So.
A
I
don't
think
I'm
looking
through
who
has
commented
here.
I
don't
think
this
is
on
the
roadmap
right
now,
I'm
paying
the
the
pm
alex.
Okay,.
D
Yeah
and
if
it's
helpful,
we
can
always,
we
can
always
do
some
testing
on
our
one
of
our
internal
environments
with
2.1
just
to
see
what
happens
but
yeah.
If
you
guys
have
any
updates.
That
would
be
super.
A
Yeah
that
so
any
testing
that
you
do
would
be
fantastic,
if
you
add
into
the
github
issue
there,
if
it's
error
logs
or
success
with
the
new
version,
that
would
be
fantastic
but
yeah.
If
you
get
some
errors,
absolutely
put
them
in
there.
It
would
be
fantastic
to
get
more
more
live
use
cases,
so
we
can
dive
deeper
into
it.
D
I
think
that
was
it
for
us.
We
can
stop
pestering
you
guys,
but
we
really
appreciate
it
and
I
echo
jeff's
statements.
This
is
really
helpful.
I'm
excited
for
the
office
hour
concept,
it's
pretty
cool.
A
Yeah
awesome
yeah
we're
we're
planning
to
do
these
at
a
recurring
on
recurring
dates.
Of
course,
we
got
a
bunch
of
holidays
and
stuff
coming
up
now
here
in
the
u.s,
so
it'll
be
a
little
broken
up,
but
then
we'll
keep
them
going
and
we'll
make
sure
we
have
even
more
people.
They
can
answer
questions
as
well.
E
Yeah-
and
I
guess
speaking
of
recurring,
this
might
be
somewhere
on
the
harper
project.
Is
there
a
general
time
frame
for
new
updates?
Like
would
2.2
be
expected
in
some
time
frame
typically,
or
is
that
just
basically
bond
feature
and
issue
completeness
the
dates
get
set.
A
So
if
you
look
at
the
road
map,
that's
that
kind
of
gives
you
a
clearer
view
of
when
certain
things
will
happen,
not
necessarily
when
a
release
will
happen,
but
rather
when
a
new
larger
feature
will
happen.
Okay,
so
that
will
give
you
a
bit
more
focus
essentially,
but
2.2
is
essentially
planned
for
for
the
end
of
the
year.
I
believe,
and
then
the
next
one
after
that
will
be
in
march
next.
C
A
And
these
are,
of
course,
tentative
dates,
and
these
are
our
wishes
on
the
roadmap,
things
that
we're
working
towards
so
yeah
future
statements
and
all
of
that
disclaimers,
but
yeah.
Those
are
the
things
that
we're
focused
on
great.
Thank
you
yeah.
I
see
we
have
a
a
new
person
joined
as
well
sparta.
A
A
So
I
have
one
question
for
for
everyone
here.
I
don't
know
if
you
have
seen
this,
but
we
do
have
I'll
share
my
screen
here,
so
you
can
all
see
it's
a
little
better.
A
We
do
have
an
adapters
md
where
we
add
in
companies
and
organizations
that
are
using
harbor
in
some
way,
and
it
would
be
fantastic
if
you
would
like
to
add
yourself
in
here.
A
You
can
see
a
lot
of
companies
here
have
added
some
success
stories
as
well,
such
as
hey
we're
running
harbor
in
production,
serving
images
here
from
china,
mobile,
serving
images
to
a
container
environment
with
over
a
thousand
nodes
and
know
about
twenty
thousand
images,
and
so
on
and
so
on.
So
if
you
would
like
to
add
your
logo
and
maybe
a
success
story
as
well,
that
would
be
wonderful.
E
Yeah
I'll
take
that
so
you
know
megan
shashank
and
I
the
company
we
work
at
is
huge
and
there
seems
to
be
lawyers
that
pop
up
all
the
time
with
various
concerns
that
I
would
never
think
of
for
putting
logos
on
pages.
But
I
will,
I
will
go
see
if
we
can
make
that
happen.
A
Yeah
that
would
be
awesome
because
we're
always
looking
for
for
more
use
cases
as
well,
and
this
is
not
just
for
for
this
particular
page
or
the
website,
but
we
also
when
we
look
at
events
going
forward
as
well.
So
looking
at
cubecon
right
now,
we
have
cuba
cubecon
here
in
less
than
a
month,
just
a
few
weeks
away
and
we're
always
looking
for
users
of
harbor
to
come
and
present
their
use
case
so
for
next
year.
If
you
would
like
to
be
co-presenting
with
the
other
harbor
maintainers.
E
E
Okay,
so
yeah
you
guys
are
at
kubecon.
I
think
several
of
us
will
be
attending
virtually
so
we'll
look
out
for
your
sessions.
A
Yeah
we're
we're
gonna,
keep
the
the
office
hours
open.
If
you
want
to,
you
can
stick
around
or
you
can
absolutely
drop
off.
That's
a
drop-in
drop-off
theme
here.
We
thank
you
all
for
your
questions
already.