►
From YouTube: Community Meeting, March 14, 2022
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
C
C
Microshift
4.8
running
on
the
Ubuntu
OS
that
supports
the
the
GPU
we
tried
to
follow.
The
steps
you
know
in
the
basic
is
easy,
quick
start,
so
we
enter
workspace
and
we
created
the
Sinker
yaml.
We
can
successfully
run
the
yaml
on
the
microchip
cluster,
but
when
we
go
and
bind
we
get
an
error
where
the
binding
is
not
taking
place,
and
it
comes
back
with
an
error
that
you
can
see
at
the
very
top.
It
says
that
The
Binding
API
cannot
bind
to
kubernetes.
C
So
since
then,
we
have
been
chatting
with
David
on
the
slack
Channel.
It
gave
us
some
suggestions,
so
a
new
command
to
run,
but
that
is
also
failing.
E
Yeah
I
mean
I
I,
think
we
we
should
maybe
after
the
Community
Color
I
had
to
leave,
for
you
know
half
an
hour
just
before,
but
maybe
we
should
look
into
why
the
the
command
I
proposed.
You
know
to
import.
E
The
resources
from
the
physical
cluster
did
not
work,
because
that
I
mean
that.
Obviously
there
is
something
wrong
here
that
would
work,
but
but
I
agree
that
it's
not
the
you
know
it's
it's
a
a
way
to
bypass
the
problem.
The
main
problem
is
that
you
know
the
physical
cluster,
the
the
kubernetes
version
of
the
physical
cluster
is
not
compatible
with
the
root
compute
apis,
and
that
raises
obviously
the
I
mean
to
fix
it
correctly.
That
raises
the
more
general
question
of
how
do
we
support
different
Cube
versions
in
the
root
compute?
E
So
the
the
there
is.
You
know
two
aspects
on
the
issue
for
the
the
one
of
the
last
command
I.
Would
you
know
come
back
to
you
to
troubleshoot
that
afterwards,
if
you
want.
C
Okay,
so
perhaps
I
can
mention
a
couple
of
things,
so,
unfortunately,
this
was
supposed
to
be
part
of
the
POC
that
we're
doing
for
Edge
sure
and
see,
and
we
are
planning
to
use
the
Jetson
Nano
because
we
use
the
same.
We
want
to
run
basically
the
same
examples
that
we
had
demonstrated
using
another.
You
know
workflow,
so
I
wanted
to
be
as
similar
as
possible.
Microshift
4.8,
it
is
my
understanding,
is
the
latest
version
that
can
run
on
Ubuntu
and,
unfortunately,
real
Forex
does
not.
E
Yeah
and
and
by
the
way,
the
last
command
that
I
proposed
to
you,
maybe
I
I
mean
I,
could
have
made
a
a
mistake
in
in
the
command
as
I
pasted
it
or
something
like
that.
So
we
just
have
to
understand
why
it
doesn't
work
in
your
case,
or
maybe
there
is
a
step
that
is
not
applied.
I
would
I
would
I
mean
turn
you
back.
E
Yeah,
that's
the
part
that
confuses
me.
Yes,
but
but
that's
not
expected.
I
mean
normally,
even
when
you
want
to.
When
you
have
some
resources
that
you
imported
from
import
from
the
physical
cluster.
The
workload
sync
command
should
work
up
to
the
end
and
you
should
be
able
to
have
the
yaml
to
apply
and
start
the
Sinker,
so
I
mean.
Obviously
there
is
one
step
which
is,
you
know
not
done
as
expected,
or
maybe
I
did
a
copy
paste
error,
or
something
like
that.
E
So
we
have
just
to
troubleshoot
that,
but
this
approach
Works
today
and
this
should
be
the
short
term
I
assume
at
least
it
could
be
a
short-term
solution
to
bypass
the
problem,
and
the
long
term
obviously
is
answering
the
question
of
how
we
manage
different
Cube
versions
in
the
in
the
cube
releases
in
the
in
the
root
compute.
F
A
Thank
you,
Franco
Andy,
you
had
your
hand
up
and
put
it
down:
you're,
good.
Okay,
all
right!
Let's
head
back
to
the
agenda,
mandiris
hope
I
pronounced
that
correctly.
I
have
questions
about
28.69.
D
Yes,
sorry
I
was
thinking,
I
would
not
be
able
to
make
it,
but
I
did
it.
So
there
was
an
open
issue
to
splits
package
client
into
separates,
go
module,
so
people
could
use
clients
without
them
during
half
of
the
world
from
kcp,
but
we
managed
it
it's
done
and
basically
the
side
effect
of
that
is
now
that
each
and
every
time
you
change
an
API
or
change
your
generation.
D
B
I
can
see
using
the
apis
without
the
generated
client,
though
so
you
know,
there's
there's
maybe
a
reason
not
to
put
them
together,
but
if
you're
pulling
in
the
apis-
and
you
happen
to
get
a
client
set-
that
you
don't
use
because
you're
using
controller
runtime,
it's
maybe
not
the
end
of
the
world.
G
So
could
we
run
through
like
what
the
actual
inconsistency
or
breakage
is?
Is
it
just
for
folks
that
are
trying
to
use
things
from
Maine.
B
Locally
would
be
fine,
it's
it's
more
of
an
administrative
challenge
once
it
gets
into
the
repo,
so
you
would
have
to
merge
a
PR
just
for
the
API
change
tag
in
git,
a
new
version
that
includes
that
API
change
then
do
another
PR
to
the
client
package
to
update
its
go
mod
to
point
at
the
just
tagged,
API
release,
and
then
you
can
generate
the
new
or
updated
clients
and.
D
D
D
H
D
B
Or
the
the
repo
needs
to
be
self-contained
for
what's
checked
into
the
branch
which
any
way
we
go
will
be
because
it
has
to
pass
CI
the
for
local
development
if,
if
I
think
it's
reasonable
to
say
if
you're
changing
the
API
and
you
want
to
update
the
client,
you
must
use
it
go
work.
B
G
B
Well
we'd
I
think
any
time
we
do
a
new
or
anytime,
we
change
the
API
module.
B
We
generally
need
to
tag
it
incrementally,
so
you
know
1201,
1202,
1203
and
so
on,
because
otherwise
there's
no
way
that
the
client
module
can
use
them.
D
And
this
is
why
I
was
thinking
about
potentially
the
iron
client
into
one
single
go
module
because
we
don't
have
a
lot
of
dependencies
together.
It
should
be
easy
to
then
the
first
terminal
project
and
in
the
long
run
long
long
run.
We
can
split
a
client
to
separate
repository
basically
and
iterate,
and
it's
individually
it's
a
bracelet.
G
G
Derivative
entirely
from
the
API
like
Auto
generated
so
I,
don't
know
that
that's
super
useful
yeah
I
mean
putting
them
together.
There's
no
impact
to
Consumers
right.
G
G
Like
if
so,
if
I'm
a
swimmer
today
and
I
I'm
using
the
generic
client
in
control,
runtime
or
something
and
I
only
want
to
learn
the
apis
with
this
change,
if
I've
been
also
pulling
the
client,
the
only
impact
of
project
is
a
more
complicated,
go
mod.
It's
not
like
any
change
to
my
build
artifacts
or
anything
like
that.
B
Yeah,
presumably
because
controller
runtime
uses
a
client
go
like
pulling
in
a
kcp
SDK
module
that
has
the
apis
in
the
client.
It's
you
know,
maybe
maybe
we
have
something
extra
in
there,
but
I
mean
our
direct
requirements.
Are
our
API
Machinery
repo,
our
client
go
repo,
our
apis
module,
our
logical
cluster
and
then
kubernetes
API
machinery
and
client
go
like
for
direct
requirements.
It's
pretty
minimal.
D
D
A
A
Okay
sounds
like:
we
have
next
steps
there,
the
next
topic
it
looks
like
Mike's
not
able
to
make
it
today
so
Andy.
You
can
speak
to
the
kubernetes
release.
B
Yeah
so
I
know
Mike
put
in
here
rebasing
on
to
125
and
126.
we're
gonna
skip
125
to.
B
Toil
and
just
rebase
onto
126.
I
about
20
minutes
ago,
did
the
first
step,
which
is
creating
the
spreadsheet
that
lists
all
the
commits
that
we
have
so
I
can
start
to
see
if
it
makes
sense
to
collapse
any
or
drop
any
and
then
I'm
going
to
start
working
on
it
now
I'm
out
next
week.
So
I
doubt
I'll
get
this
done
by
Friday,
but
I'll
at
least
get
it
started
this
week.
A
Okay,
sorry
Mike
audio
cut
out.
Do
we
want
to
talk
about
the
next
kcp
release
before
we
look
at
the
playground,
plugin
I
know.
We've
talked
a
little
bit
about
what
we
want
to
put
in,
but
I
don't
know
if
we've
talked
about
it
in
the
community
meeting.
H
B
It-
and
that
was
on
purpose
so
that
we
didn't
just
keep
punting
from
one
Milestone
to
the
next.
So
there
are
over
300
issues
that
are
currently
open,
so
I
think
it's
up
to
us
as
a
community
to
decide
what
major
feature
or
features
we
want
to
try
and
work
on
for
0.12.
G
Sometime
in
the
distant
past,
we
used
to
do
a
release
per
month.
Do
we
have
an
idea
that
we
want
to
get
back
to
that
Cadence
like
is
this
next
one?
Would
we
like
to
bring
it
down
at
the
end
of
the
month.
B
I
I
don't
know
how
much
has
gone
into
the
main
branch
since
our
last
release.
I'm
not
opposed
to
trying
to
do
one
at
the
end
of
the
month.
If
it
makes
sense
or
we
could
try
and
do
it
sometime
in
April,.
G
Okay,
and
do
we
and
maybe
Mike,
can
speak
to
this?
Do
we
have
an
idea
if
the
rebase
is
going
to
bring
us
like
things
that
we
really
need
should
that
be
in
there.
B
I
have
not
looked
that's
another
part
of
the
rebase
steps
is
to
look
at
the
packages
that
make
up
generic
control
plane
from
the
their
sources
and
see
what
the
dips
were
between
124
and
126.,
and
then
I
know
that
there's
things
that
are
in
the
main
branch
that
didn't
make
it
into
126
because
they
came
in
after
that
would
be
nice
to
have
the
contextual
logging
changes
for
the
quota
and
garbage
collection
controllers
would
mean
that
we
generally
either
don't
have
to
carry
patches
or
the
patches
get
a
lot
smaller,
but
that's
not
in
I
could
try
to
backboard
them.
B
That
might
help
and
I
know
that,
like
yeah
I
have
to
check
so
I,
don't
know
off
the
top
of
my
head.
It.
A
G
Let's
open
it
discussion
for
nominating
things
to
go
into
the
next
release
and
then
we'll
figure
out
if
the
timeline
for
the
month
looks
reasonable,.
A
Thank
you
and
then
let's
look
at
this
playground.
Plugin
I,
don't
know
if
Mike
made
a
comment
in
here.
F
F
A
Yeah,
it
looks
like
Mike
has
some
questions
for
Fabrizio,
possibly
but
I,
don't
think
Fabricio
is
here
so
I'm,
not
sure
what
more
we
can
do
to
answer
this
question
for
the
moment
and
finally,
interface
for
identifying
Associated
objects
for
upsync.
Does
anyone
have
contact
on
that.
B
E
Mean
sorry
yeah,
maybe
just
just
one
point
here:
probably
it's
even
you
know
not
that
complicated.
Currently
the
resource
types
that
can
be
absinct
is
fixed
and
hard
coded
into
the
sinker.
E
So
there
is
not
not
even
you
know
a
customization
where,
or
you
know,
a
declarative
way
to
define
that
we
want
to
add
some
new
resource
types
to
be
absinct,
and
so
that's
the
first
level
of
of
you
know
being
able
to
drive
the
what
the
object
types
should
be
absinct
and
then
the
second
level
is
the
one
requested
by
I
think
Paulo
several
times,
which
is
decide
what
we're
seeing
at
the
instance
level,
and
especially
if,
if
the
I
think
ibm1
or
some
contributors
wanted
to
absync
some
secrets,
for
example,
from
Downstream
to
to
Upstream
so
I
mean
there,
we
I
mean
everything
is
to
to
be
built
here.
E
E
You
know
critical
informations
from
Downstream
to
Upstream,
so
I
mean
it
seems
to
me
that
there
there
is
a
full
design
here
too,
to
invest
in
before
being
able
to
you
know,
have
this
Landing
in
in
some
release
for
now
at
thinking
is
really
really
just
hard-coded
for
the
very
minimal
use
cases
that
were
requested
until
now,
which
is
upsyncing
Pods
at
syncing
at
points
and
possibly
appsyncing
I
mean
soon
PVS,
for
you
know,
storage
movement,
but
that
was
very
low
level
and
and
standard
Cube
related
hard
cut
it
into
the
Sinker.
F
A
Yeah
I've
seen
other
projects
also
list
like
these
are
things
that
use
it
or
here's
other
use
cases
things
that
use
it
as
a
dependency.
That
sort
of
thing.
F
B
I
also,
in
addition
to
make
that
repo
I
just
made
a
friends
repo-
and
we
had
talked
about
this
I-
don't
remember
if
it
was
at
a
community
meeting
or
around
slack
or
somewhere,
but
other
other
repos
have
a
a
friend's
repo
where
folks
can
come
in
and
say
who
they
are
and
how
they're
using
the
project
so
I
think
that
would
be
a
cool
thing
to
do
as
well.
I'll
put
together,
read
me
or
if
somebody
wants
to
get
filmed
before,
I
do
for
these
two
new
repos
that'd
be
cool.
F
B
B
Yeah
six
door
has
one
you
want
to
look
at,
say,
store,
slash,
friends,.
B
And
I
would
say
you
know
you
could
just
pick
any
one
of
the
ones
in
there
at
random
and
just
open
up.
You
know
so
people
can
see
what
one
of
these
looks
like
you
know,
it's
not
a
lot
or
it
doesn't
have
to
be
a
lot.
It's
just
one
or
more
short
sentences
about
how
they're
using
tecton
or
six
door
or
whatever.
A
Okay,
cool
well,
those
are
out
there
for
now
we'll
you
can
get
some
readme's
and
maybe
some
templates
put
in
small
templates
all
right.
We
have
29
minutes
left
on
the
agenda.
Does
anyone
have
anything
the
last
minute
they
didn't
put
in
the
issue
that
they'd
like
to
bring
out
before
we
go
to
triaging.
D
B
A
E
Yes,
future,
you
probably
yeah
you
can
put
that
in
next.
Okay,.
E
H
G
A
G
A
That
makes
sense:
okay,.
E
Maybe
we
can.
We
can
comment
that
in
the
in
the
issue
and
and
let
the
discussion
start.
Obviously,
that's
not
something
short
term
and
yeah.
A
Okay
bug
we
use
disgust
and
yeah.