►
Description
CNCF Harbor's Community Zoom Meeting
A
Okay,
what's
up
hello,
everyone
welcome
to
the
first
of
may.
The
first
community
meeting
in
may
today
is
may
the
fourth
and
happy
star
wars
day.
Everyone
first
of
all
and
second
of
all,
welcome
and
please
keep
in
mind.
That's
the
official
community
meeting,
so
please
behave
follow
the
code
of
contact
and
overall
be
a
nice
person.
My
name
is
olympus
silith,
I'm
located
in
bulgaria.
I'm
saying
that
because
I
can
see
some
new
folks
in
the
call
I've
I've
pasted
the
community
meeting
link
for
the
notes.
A
So
if
you
feel
like
adding
yourself
to
the
attendance
list,
I've
changed
that
a
bit
today,
I'm
going
to
share.
A
So
I
I
changed
the
format
a
little
bit,
so
we
can
have
it
more
structured.
I
took
that
from
the
valero
project.
I
I
I
see
that
that
works
there,
because
we
can
see
who
attended
and
if
there
is
a
point
for
someone
who
wasn't
there,
it's
easy
to
to
address
them.
So
I've
changed
the
plan
a
bit
so
yeah,
that's
the
new
one
compared
to
the
to
the
old
one
over
here,
and
so
please
add
your
topics.
A
If
you
have
any,
if
you
don't
have
any
feel
free
to
introduce
yourself
right
now,
I
can
see
mark,
and
I
can
see
I'm
sorry,
but
I
don't
know
how
to
read
that
dior
guerra
is
it?
Is
it
right?
I'm
sorry
for
my
prince,
hey,
okay,
so
I
have
two
things
for
today
feel
free
to
add
yours.
First
of
all,
I
wanna
recognize
the
the
the
pr's
from
simon,
I
suppose,
is
his
name.
A
He
has
contributed
quite
a
lot
in
the
last
few
periods
and
especially
around
the
ui,
which
leads
me
to
my
next
topic
is:
if
you
can
take
a
look
at
the
refactoring
of
the
portal
language
code,
I've
asked
the
maintainers
to
to
review.
That
code
looks
pretty
good
for
me,
but
I'm
not
the
expert
here
so
young
one
and
the
rest
of
the
team.
Please
take
a
look
at
this
one
and
give
your
feedback
looks
pretty
good.
It
will
make
the
localization
a
bit
more
easy
to
swallow.
A
A
I've
ordered
all
the
slacks
and
all
the
goodies
that
we're
gonna
give
away
to
the
community
or
whoever
wants
to
show
up
at
our
kiosk.
I
think
I'm
one.
You
have
recorded
your
maintenance
track
already
right.
A
Radio
record
nice,
so
it's
already
up
there
and
ready
to
to
be
streamed
right.
Yeah,
nice,
nice,
nice,
okay,
okay,
yeah!
I
can
update
you
on
a
few
other
stuff,
the
stale
pr
that
have
opened
last.
I
don't
know
when
yeah
almost
a
month
ago
it
was
merged.
So
I
suppose
in
20
20
days
we
will
start
having
results
after
this
one.
So
we
can
get
rid
of
super
old
stales
and
you
can
see
the.
A
A
B
A
Someone
opened
it
too
long
ago,
yeah
three
on
march
march
3rd.
No,
since
then,
okay,
I'm
going
to
remove
this
tail
from
this
one.
I
really
hope.
B
We
will
resolve
this
problem
in
2.6.
B
A
A
Let's
have
some
activity
here
not
to
get
the
stereo
again,
okay,
so
with
that,
I
I
don't
have
anything
else.
If
you
want
to
add
something
or
you
want
to
ask
questions
now
is
the
time.
C
So
I
edited
the
discussion
topic,
hey
guys
mark
here.
There
is
an
issue
I
opened
a
while
ago
in
the
hardware
operator
repository.
C
C
C
Yeah,
so
right
now
the
problem
actually
goes
back
to
the
docker
registry,
the
upstream
docker
registry
project.
C
It
doesn't
work
with
with
the
latest
workload,
identity,
authentication,
solutions
from
aws,
and
for
that
reason
we
need
to
use
access
key
and
secret
key
to
authenticate
towards
s3
and
the
way
it's
currently
configured
in
the
hardware
operator
or
in
the
custom
resources
is
that
you
set
a
secret
kubernetes
secret
reference
for
the
secret
key
from
the
from
the
credentials,
but
you
actually
have
to
paste
the
access
key
part
in
the
custom
resource,
which
is
a
bit
problematic
if
you
want
to
automatically
rotate
those
credentials
so
right
now,
if
you
want
to
rotate
those
credentials,
you
have
to
manually
update
the
access
key
part
in
the
custom
resource,
which
again
is
a
bit
problematic.
C
If
you
do
githubs,
because
then
you
have
to
go
in
and
manually
update
the
the
custom
resource
in
githubs
and
then
you
can
update
the
secret
separately,
so
I
propose
a
couple
different
solutions
there.
Most
of
them
include
moving
the
access
key
to
either
separate
secret
or
to
the
same
secret
as
the
as
the
secret
reference
for
the
secret
key.
C
But
really
the
goal
here
is
to
move
the
access
key
out
of
the
custom
resource
to
a
secret
where
we
can
automatically
rotate
both
at
the
same
time
using
some
other
tools
so
personally
or
or
at
cisco.
We
use
external
secrets
to
synchronize
secrets
from
hashicorp
vault
to
kubernetes
and
we
use
hashcardboard
to
issue
aws
credentials.
C
C
C
I
can
see
that
it's
scheduled
for
release
1.4
right
now,
which
is
two
releases
out,
I'd,
be
happy
to
work
on
if
that
speeds
up
the
process,
because
this
is
this-
is
this-
is
critical
for
us
because
right
so
we
we
have
a
policy
to
rotate
every
secret
every
60
days
and
doing
that
manually
is
a
is
a
bit
of
a
pain.
So
we
we
try
to
make
that
automatic
as
much
as
we
can.
A
Yeah,
if
you
can
work
on
that,
it
would
be
super
cool
steven
zu
is
your
person
of
contact
for
the
operator
decisions
and
and
stuff.
A
So
if
you
have
something
prototyped
or
if
you
have
time
to
to
work
on
this,
one
get
in
touch
with
him
or
this
issue
or
if
you,
if
you
have
already
something
working,
you
can
pr
it
and
and
then
just
trigger
a
discussion
and
yeah.
C
There
is
the
reason
I
haven't
really
started
working
on
it,
because
there
are
a
couple
different
ways.
We
could
solve
this
problem
so
first
I
wanted
to
make
sure
that
we
settled
on
at
least
one
and
then
maybe
provide
a
pr
so
solution.
Part
itself
is
not
complicated,
in
my
opinion,
finding
the
right
solution,
that's
both
backward,
compatible
and
yeah.
It
makes
sense
for
from
the
problem
perspective.
That's
that's
kind
of
the
first
step.
C
A
Yeah
yeah,
can
you
say
why
it
was
implemented
for
the
last
20
days,
but
as
soon
as
I
think
january?
How
long
is
your
holiday
in
in
china
until
next
week,
right.
B
It's
just
till
today.
A
Yeah
yeah,
okay,
so
I
suppose
in
that
case
stephen
will
respond
in
the
next
few
days.
A
C
That's
great
thanks
and
by
the
way
operator
do
you
know
when
the
next
batch
release
will
be
released,
because
I
have
a
patch
merging
into
the
1.2
release
branch.
But
I
haven't
seen
the
release
yet
so.
A
I'm
not
so
much
into
the
the
operator,
so
I
cannot
okay.
I
cannot
say
right
now,
but
I
I
can
check
that
for
you
and
and
think
your
slack
or
something
about
this
one.
Okay,.
A
That
should
be
somewhere
here
either
yeah,
I'm
sorry!
You!
You
catch
me
unprepared
for
this.
For
this
for
this
question,
but
I'll
try
to
find
out
the
information
you
need
yeah.
C
I've
just
so
really
I'm
interested
in
general.
What
the
release
cycle
looks
like
for
the
operator,
including
both
minor
and
patch
versions,
because
yeah,
as
I
said
there,
I
think
there's
only
one
in
the
in
the
current
patch
branch
right
now,
but
yeah
I'm.
So
if,
if
he
could,
I
don't
know
it
might
make
sense
to
to
document
it
somewhere
in
the
redmi
or
somewhere
on
the
repo.
A
Yeah
I'll
add
that
to
the
to
the
community
meeting
that
we
we
have
to
work
on
proper
roadmap
anytime
timelines
for
the
operator
as
well.
I'm
not
sure-
maybe
maybe
that
that's
there,
but
I
have
to
check.
C
Yeah,
generally
speaking,
for
for
patch,
I
think
bachelor
of
this
should
be
fairly
easy
to
release,
because
those
are
generally
just
bug
fixes
and
usually
there
are
matrixes
people
are
worth
waiting
for,
but
yeah
for
minor
releases.
A
Yeah,
okay,
I'll
try
to
catch
stephen
tomorrow
and
see
because
he's
he's
the
driving
force
behind
the
the
operator.
So
he
can
I'm
sure
he
knows
the
answer.
I
I
just
don't
know
it.
Okay,
I
don't
have
a
right
yeah.
C
And
just
for
you
to
know
so
right
now
we
are
relying
heavily
on
the
operator
because
we
managed
to
few
different
hardware
instances
and
we
actually
kind
of
released
hardware
as
a
managed
service
internally
at
etni.
So
different
teams
can
ask
us
to
to
spin
up
new
hardware
instances
and
we
use
the
operator
for
that
to
manage
different
hardware
instances.
C
So
that's
that's
why
the
the
operator
is
critical
for
us
just
for
you
to
know.
A
A
Okay,
you're
the
same
mark.
Okay,
then.
Yes,
I
am
okay,
yeah
okay,
so
I
can
follow
up
on
the
release
stuff
and
also
I
wanted
to
discuss
with
you
how
cisco
utilizes
harbor,
if
we
can
do.
C
Yeah,
of
course
I
mean,
I
think
I
think
we
actually
have
a
pretty
interesting
use
case
because
at
ethiopia,
ethnic
is
a
business
unit
within
cisco,
where
we
have
a
bunch
of
different
project
teams
that
require
cardboard
for
their
their
own
projects.
But
we
also
use
hardware
as
a
means
to
distribute
artifacts
to
customers
and
we
actually
hit
a
couple
issues
there.
So
maybe
we
could
talk
about
those
especially
around
the
replication
performance,
but
I
think
we've
come
up
with
an
interesting
setup
for
distributivity
artifacts
to
two
customers.
A
Okay,
if
you
can
check
on
your
site,
how
much
you
can
disclose
will
be
great,
of
course,
yeah.
So
we
can
yeah,
we
can
talk
about
it
and
like
formalizing
it
in
some
kind
of
blog
post
or
something
like
this
will
be
super
cool.
D
About
the
replication,
well,
we
actually
yesterday
had
an
issue
where
we
we
set
up
a
replication
job
with
50
000
objects
and
it
kind
of
broke
our
production.
D
D
And
what
can
we
do
here?
Because
it
was
quite
hard
for
us,
because
we
didn't
know
how
how
the
replication
was
working
so
discovering
going
through
the
database
and
deactivating
radius
so
that
we
could
actually
stop
this.
So
we
we
can
have
our
infrastructure
working
with.
By
also
is
there
any
provision
of,
because
we
feel
that
the
jobs
should
run
in
background
right
and
the
this
should
be
not
noticeable.
B
Could
you
just
tag
me
in
the
in
your
abstract
so
that
we
can?
I
can
respond
to
your
a
problem
and
and
by
the
way,
what's
the
version
of
arbor?
Sorry,
what's
the
worst
of
harvard.
B
Yeah
it's
two
of
two
for
two
yeah
yeah
right.
So
please
just
check
me
in
the
section
also
that
I
can
and
have
a
response
to
your
problem.
C
A
C
Something
that
just
we
just
solved
with
adding
more
cpu
to
the
nodes
for
now.
So
that's
fine,
the
other
a
little
bit
bigger
issue
that
we
recently
quite
recently.
Actually
so
it's
like
from
last
week
so
was
that
during
the
applications
when
there
was
a
lot
of
things
to
replicate
the
the
registry
with
the
registry
experience
some
some
connection
drops,
which
is
quite
unexpected,
and
I
can't
say
for
sure
that
it
was
because
of
the
registries,
but
but
it
was
definitely
occurring
during
the
replication.
So
this.
B
Now
the
first
one
for
the
cpu
usage,
we
have
faced
some
problem
or
not.
You
probably
can
upgrade
hardware
to
2.5.
So
then
how
much
yeah.
C
So
yeah,
so
we
are
still.
We
are
still
on
the
the
1.1
version
of
the
operator,
which
comes
with
2.3
version
hardware,
so
you're
still.
C
Pretty
old,
I
know
I
know
that
we
are
working
on
upgrading
to
1.2
of
the
operator
that
comes
with
to
2.4
hardware.
Hopefully
there
will
be
a
release
of
the
operator
that
supports
2.5
as
well.
B
Yeah,
that's
the
cpu
yo.
This
probably
has
already
fixed
in
the
2x5,
so
you
just
control
that
portion
and
and
for
the
second
one
and
have
you
ever
five
problem
or
in
the
github
or
or
not.
C
Once
I
have
more
information,
I
will
open
an
issue
and
and
again
we
are
still
on
an
hardware
version
so
that
that
could
be
the
problem
as
well.
B
Okay,
okay,
so
if
you
find
anything
just
send
message
to
success
channel
and
then
I
can
have
a
look.
A
Okay,
if
there
are
no
more
topics
to
discuss
once
again,
I
want
to
bring
the
might
ask
for
everyone.
If
you're
planning
to
visit
kubecon,
stop
by
the
eu,
so
buy
the
kiosk
or
yeah
just
try
to
say
hi.
I
want
to
meet
everyone
who
is
part
of
the
community
and
also,
if
you're,
planning,
to
visit
kubecon,
north
america
and
you're
somehow
feeling
chatty,
and
you
want
to
do
a
hardware
presentation
or
any
kind
of
talk
about
harbor.
A
Also,
please
contact
me
I'll
try
to
help
you
out
with
this
with
the
stuff,
via
the
all
the
cncf
channels
and
to
to
formalize
your
aspect
and
everything
so
yeah.
A
A
Okay,
without
if,
if
that's
all,
once
again,
happy
star
wars
day
enjoy
geek
out
and
see
you
next
time,
then.