►
From YouTube: CNCF Harbor's Community Zoom Meeting - November 16, 2022
Description
CNCF Harbor's Community Zoom Meeting
A
There
we
go
hello,
everyone,
today's
November
16th
and
does
the
official
community
meeting
for
Harbor.
My
name
is
Olinda
Silva
and
I'm,
the
community
manager
for
Harbor.
So
this
this
official
community
meeting,
so
please
follow
the
code
of
conduct
and
practically
be
nice
to
each
other.
I'm
gonna
share
the
screen.
If,
if
someone
added
something
for
today,
let's
see
no
all
right,
also
I'm
gonna
paste
in
the
chat.
Once
again,
the
link
to
to
the
community
meetings
notes.
A
Oh
nice,
nice
cool,
okay,
great
I,
can
see
some
new
faces
on
on
our
call.
Can
you
introduce
yourselves
please.
B
So
I
can
start
and
myself
I
work
at
Giant
swan
I'm
here,
together
with
Will
and
Depot,
we
work
on
our
project
together
to
run
Harbor,
basically
on
our
kubernetes
clusters
and
there's
a
few
topics.
We
would
like
to
discuss
so
I'm
going
to
note
them
down
and
yeah.
Maybe
we'll
end
the
book
and
introduce
some
stuff
very
quickly
as
well.
C
Yeah,
hello,
everyone,
my
name
is
I
work
with
container
Solutions
and
currently
working
with
myself
on
and
will
and
it's
upper
on
a
chance
on
project.
So
next
week.
D
Yeah,
hey.
D
Your
wife
just
disappeared,
I
think
it's
the
Sun,
but
I
am
I,
will
not
not
to
be
confused
with
the
tree
and
yeah
I'll
work
with
marceland
Depot
on
this
on
this
project
together.
A
Nice
is
that
the
same
project
that
you
share
with
Gustavo
from
container
Solutions,
yes,.
A
Gustavo
and
Ben
they-
they
talked
with
me
already:
yeah,
okay,
nice.
So
if
you
feel
like
it
or
we
can
take
that
or
find
and
chat
a
little
bit
about
the
project
and
how
you
plan
to
implement
Harbor,
we
can.
We
can
write
a
like
a
use
case
or
some
some
kind
of
blog
about
this
one.
If
you're
feeling
like
open
to
disclose
what
you're
gonna
do
and
the
setup,
it
would
be
super
nice
for
the
rest
of
the
community
to
see
how
you're
going
to
do
that.
B
A
Okay,
since
that,
let's
first
okay,
I
I
see
Stone
out
at
the
job
service
demo
for
today.
Do
you
mind
if
we
do
that
demo
first
and
then
we
can
discuss
all
the
rest
of
the
topics.
Sure
no
rush,
okay,
okay,
Stone!
The
floor
is
yours.
Take
it
away.
E
Okay,
let
me
share
my
screen.
E
E
Okay,
let's
talk
about
the
trapezo
is
let's
Bother.
The
episode
is
our
foreign
work
to
schedule
a
different
task
in
async
mode,
such
as
a
replication,
garbage
collection
and
images
and
type
of
retention.
Etc
sometimes
have
a
user
might
have
found
that
their
replication
does
not
complete
in
time,
and
there
is
something
wrong
with
the
replication
of
the
the
replication.
Java
is
always
in
Penny
status.
So
it
is
a
very
difficult
for
user
to
Diagnostic
and
mitigate
a
recover.
E
E
It
should
be
the
administrator
to
log
in
to
have
to
use
this
feature.
E
There
is
a
new
money
item,
job
service,
dashboard
and
the
job
service.
Dashboard
has
three
three
things
to
three
components:
the
job,
Java,
Q
schedule
and
worker.
This
is
the
dashboard
Matrix
and
under
this
dashboard
there
it
is
their
detail,
information
about
the
job,
queue
and
schedule,
and
workers.
E
The
first
planner
is
the
job
queue
status.
We
could
check
the
account
job
queue
status,
for
example
the
Cabbage
collection
and
image
Scan.
They
have
different
job
queue
for
replication,
they
have
a
duplication,
queue
for
slack
and
they
have
they
have
a
slightly
job
queue
and
there
is
a
pending
conduct.
The
pending
counter
is
the
total
jobs
in
in
a
queue
and
the
latency
is
the
max
the
maximum.
E
What
it
has
if
you,
if
the
counter
latency
is
one
hour,
it
means
that
your
job
should
take
one
hour
to
get
executed
and
the
last
one
is
the
is
pause
the
status
when
it
is
paused
the
job,
the
job.
We
are
not
a
executed
until
when
we'll
stop
to
pick
up
from
the
queue,
and
until
we
resume
it
from
the
resume
the
queue
then
it
will
get
executed.
E
The
second
type
is
the
schedule.
There
are
a
lot
of
jobs
run
as
a
scheduled
task.
For
example,
the
damage,
collection
and
replication
can
be
scheduled
with
our
daily
or
weekly
schedule.
Is
that
we
can?
We
can
specify
our
frequency
to
run
a
job
actually,
this
job,
this
scheduled
job,
is
a
and
has
a
job
tab.
This
job
type
is
the
scheduler
all
this
job.
All
these
schedules.
E
E
The
last
of
the
panel
is
the
worker
The
Walker
poor.
Currently,
there
is
only
one
worker
poor,
because
we
start
this
Harbor
without
compose.
When
you
are
running
in
kubernetes,
you
can
specify
multiple
job
service
part.
So
this
will
create
many
multiple
worker
ports
for
each
course.
It
will
create
a
money
and
workers,
they
have
main
workers.
Currently
there
are
10
workers
running
in
a
work
pool
and
there
is
no
job
in
the
running
status.
E
We
can
click
free
to
free
the
accountant
running
job
in
a
workout.
No,
let's
take
an
example.
Let
me
start
a
replication
image
replication.
This
is
this:
is
our
application
we
have.
We
can
replicate
a
almost
50
images
and
we
have
a
limited
bandwise
to
replicate.
We
have
limited
bandwise
to
10
000.,
one
thousand.
E
E
It
will
remove
all
the
pending
jobs
in
the
replicating
Java
queue
and
the
market
displays
at
the
start
and
as
we
know,
as
we
just
discussed
there
are,
some
tasks
is
already
in
running
status,
so
we
are
still
in
a
running
in
the
world
of
war.
So
we
can.
We
have
to
click
the
free
button
in
a
worker
pool
it
will
clean
up.
It
will
stop
all
that
running
tasks.
E
Let
me
have
some,
but
in
a
kind
of
demo
environment
it
is
fair
to,
as
we
have
I
think
we
after
we
click
the
free
button.
This
job
ID
should
be
cleaned
up
from
worker.
E
E
E
In
some
cases,
some
users
remove
that
replication
Java
already
triggered,
because
it
will
cause
that
they
had
inconsistency
between
the
two
Hardware
instances,
for
example
there
how
to
replicate
the
master
node
to
our
Dr
Hubble
instances,
if
you,
if
you
if
there
are
if
this
replication
tasks
are
removed
from
the
queue,
then
that
video
calls
the
data.
Inconsistency
between
the
you
must
know
that
under
the
Dr
side,
so
they
just
wanted
to
job
service.
Do
not
spend
too
much
time
too
much
resource
in
the
location
I.
E
E
One
that
when
you
pause
the
replication
double
Q,
you
still
can
submit
the
job
to
this
queue.
For
example,
I
could
look
trigger
this
replication
again
replicate
the
lubrication
again.
E
F
Yeah
I
think
only
is
dropped
due
to
some
images.
Let
me
continue
so
next
I
think
yeah
only
I
should
have
update
regarding
the
Auto
Sani
for
the
new
PR
in
the
harbor
main
repo.
The
idea
is,
we
want
the
P
once
the
pi
submit
to
the
repo
we
wanted
to
be,
the
reviewer
could
be
assigned
so
which
can
be
engaged
by
the
reviewer
immediately.
F
G
B
So
the
first
topic
is,
we
had
some
troubles,
so
we
are
users
of
our
Harbor
operator
right
now
to
manage
Harbor
and
one
part
of
it
is.
We
are
using
the
provided,
postgres
SQL,
that's
managed
by
a
salando
operator,
so
Solano
postgres
operator,
and
we
had
some
serious
trouble
getting
any
backups
of
the
postgres
SQL
database
working.
B
So
basically,
as
far
as
I
understood
and
correct
me,
if
I'm
wrong
or
we
understand,
Harbor
uses
a
SQL
database
to
essentially
store
data
and
index
of
yeah,
whatever
actual
images
and
so
on,
are
being
stored.
If
the
database
is
lost,
it
cannot
be
rebuilt.
So
if
we
reduce
this
database,
we
are
in
trouble.
Therefore,
it's
quite
important
for
us
to
have
backups
and
be
able
to
restore
from
those
backups
in
these
cases,
since
we
need
to
run
Harbor
on
premise,
so
we
can't
use
the
managed
solution
for
the
database.
B
According
to
this
there's
currently
two
open
pull
requests
open
by
Will
and
I.
Try
to
look
a
bit
into
Hardware
operator
and
I'm
a
bit
confused
number
one.
How
we
should
proceed
so
the
pull
requests
are
open.
Do
we
need
to
Ping
maintain
us
as
reviewers
on
the
pull
requests,
since
the
documentation,
I
found
was
was
very
out
of
date.
There
seems
to
have
been
like
a
working
group
for
a
harbor
operator,
but
the
last
meeting
I
can
find
is
I,
can
May
2021
or
something
like
this?
B
So
it's
hard
for
me
to
judge
what
we
need
to
do
to
move
forward
here.
The
the
pr
was
only
opened
recently
I'm,
mostly
asking
to
understand
the
process
correctly.
F
Yeah
I
think
currently
in
this
meeting.
Actually
we
do
not
have
a
maintenance
from.
H
We
had
a
meeting
to
review
players
and
issues
that
yeah
and
they
do
not
have
a
lot
of
time
to
to
the
operator
stuff
because
they
are
not
really
focused
on
Harbor
community
and
so
yeah.
That's
correct.
The
the
documentation
is
out
of
that
on.
H
We
know
that
we
we
build
proof,
pay
more
attention
on
that,
but
it
still
depends
on
the
the
afro
from
the
operator,
Community
Nintendo
group
and
yeah.
Every
London
knows
about
this
part
and
we
have
a
stock
channel
so,
and
you
can
also
ask
questions
about
operator
in
the
CNC
of
Hardware
section
also
and
then
I
can
assign
the
containers
from
yeah
from
operator
group.
H
G
H
You
can
just
yeah
yeah
I
got
you,
you
can
just
send
your
message
into
this
since
Samsung
Channel,
we
had
an
engineer
who
regularly
or
monitoring
the
United
States
response
every
task
I
required.
Okay,.
B
I
will
send
a
message
in
a
section
hover
operator
def,
with
both
of
the
pull
requests
that
are
open,
yeah.
H
B
H
F
So
Michelle,
which
fear
you
are
talking
about,
is
this
yeah.
B
These
two,
because,
basically
it's
not
clear
because
currently
no
reviewers
are
signed
and
if
you
look
at
the
older
requests,
a
lot
of
them
only
have
traction
from
the
Creator
itself
and
most
of
the
close
pull
requests
were
also
closed
from
the
people
themselves.
So
that's
why
I
was
asking
what
what
the
general,
how
it
generally
Works
in
this
project,
I'm.
H
Just
to
have
a
follow-up
question
on:
let's
see
so
we
you
guys
wanted
to
join
the
the
I
mention
their
group.
I
mean
you
want
to
take
some
time
to
help
us
to
review
the
PRS
features
and
we're.
B
Like,
if
that's
a
possibility,
sure
I'm
open,
because
that
would
make
our
life
easier
right.
B
Doing
is
right
now
we
need
to
upgrade
the
speedo
version
and
that's
related
to
some
discoveries
we
made
for
basically
recovery
and
then
the
second
pull
request
is
also
changing
some
variables.
If
we
can
become
a
maintainer,
then
then
sure
there's
also
my
second
point,
which
is
a
larger
change.
That
I
just
wanted
to
briefly
discuss
here
that
all
in
no
support.
So
my
second
point
is
like
managing
how
about
configuration
itself
through
an
operator
right
now.
B
It
manages
spinning
up
database
or
Mineo
or
something
like
this,
but
it
does
not
manage
the
hardware
configuration
itself
and
harbor
can
only
be
configured
through
either
the
UI
or
yeah,
the
UI
or
through
the
API,
but
it
cannot
be
configured
through
an
operator
interface
which
would
be
important
to
us
and
it's
something
we
are
working
on
right
now
to
have
our
own
operator,
which
would
reconcile
a
custom
resource
and
then
do
calls
to
the
harbor
API
to,
for
example,
create
a
new
project
set.
G
B
H
Yeah,
so
you
can
so
I
think
I
will
talk
to
the
Nintendo
from
the
operator
group
so
that
we
can
have
more
discussed
in
the
sub
Channel
about
more
in
the
future.
How
we.
H
B
Not
sure
if
you
go
to
the
Hub
operator,
maintainer
group,
there's
quite
a
few
maintainers
and
I
wasn't
sure
like
was
actually
active,
photo
ping
and
so
on.
G
F
Yeah
I
think,
if,
if
you
guys
can
turn
you
a
bit
more
on
the
Hub
operator
and
I,
think
you
do
a
great
either
for
the
project
itself
and
also
for
Community
yeah
I.
Think
we
we,
we
I,
think
it's
a
project.
Is
a
community
project
right.
So
yes,
actually
it
will
come.
B
Then
I
will
make
sure
that
we
write
these
things.
Also
in
the
slack
Channel
more
and
then
ping
you
in
case
we
don't
get
a.
B
And
we
have
talked
to
Allen
not
directly,
but
through
co-worker,
of
container
solutions
that
has
talked
to
online
as
well,
and
that
way
we
can
also
keep
communication
going
well
that
works
for
me
for
now.
F
F
Okay,
great
so
we'll
give
half
an
hour
back
to
all
the
attorneys.
Thank
you
all.
Thank
you
for
joining
the
community
meeting,
see
you
next
time.