►
From YouTube: KubeEdge Community Meeting 20200421 (Pacific Time)
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
B
Right
the
recording
is
on
so
welcome
everybody
to
the
cube
edgy
community
meeting
today
is
april
22nd,
it's
pacific
time
meeting
today.
So
we
have
a
few
items
on
the
agenda.
B
B
And
also
please,
everyone
on
the
meeting
leave
your
names
in
the
attendees.
A
B
B
B
So
before
going
through
the
future
details
that
I
want
to
highlight
that,
so
we
are
still
targeting
code
freeze
by
end
of
april,
so
it's
approximately
10
days
left.
B
And
so
generally
I
mean
previously,
we
are
targeting
the
two
weeks
after
could
freeze
for
the
for
the
1.3.0
release,
as
the
may,
as
it's
going
to
be
may
day
festival.
Shall
we
make
it
one
one
more
week.
C
Let's
targeting
try
to
do
the
middle
of
may,
let's
see
we,
we
could
postpone
right,
but
but
I
don't
know
it's
what
is
april
30th,
what
time
is
china
time
or
pdt?
That's
a
half
day
almost
a
day
different.
We
have
18
hours
a
16
hour
difference.
B
So
probably
it
may
be
the
china
time
end
of
april
30
because
we
need
to
so
by
the
code
phrase.
We
need
to
have
a
beta0
release.
C
B
Updates,
so
let
me
check
if
dave
is
on
the
meeting,
maybe
not
so,
for
the
cryo
integration
recording
the
status
update
from
dave
last
week.
The
the
changes
has
been
done
so
so
I
think
you
only
need
need
some
document
to
clarify
how
to
make
it
happen.
B
Okay,
so
so
the
the
cut
up
antenna
integration
is
original
is
actually
not
in
the
one
door
three,
but
it's
kind
of
in
parallel
task.
So
I
think
no
updates
yet.
B
So
I
think
everybody
can
start
doing
some
review
first,
while
it's
still
under
development,
otherwise
are
not
able
to
get
it
in,
but
next
week.
I
think
because
our
review
still
takes
some
time.
B
Okay
and
the
logs
and
the
exec
john
j,
would
you
like
to
give
updates.
D
B
B
Or
maybe
we
can
try
make
sure
a
lot
of
logs
can
get
in
this
way.
This.
B
C
B
So,
actually,
I
think
there
are
some
risk
of
the
supporting
efcc,
because
currently
the
difference
with
the
logs
and
the
exec
is
that
exec
need
spdy
protocol
support
and
currently
I
think,
jai
have
not
yet
finished.
B
So
I
mean
exec:
maybe
we
are
not
able
to
get
it
down
into
this.
A
C
B
C
We
split
into
two
one
in
one
point:
three,
one
in
one
point
four.
B
Yeah,
I
think
I
think,
with
this
we
we
should
make
sure
the
logs
support
getty.
B
Okay,
so
the
matrix
support-
I
think
this
is
already
in
and
we
just
need
updates
on
the
document.
B
B
Okay
and
for
gateway
support
so
currently,
actually
there,
there
there's
some
code
under
development,
it's
in
in
the
branch
here
and
the
pull
request
may
be
submitted
by
today
or
tomorrow
and
if
any
one
of
you
are
interesting,
maybe
you
can
first
of
all
go
to
the
branch
to
take
a
look
and
start
review.
B
B
This
one
we
we
have,
we
now
have
a
working
in
progress,
pull.
B
E
Now
we
have
a
pi
to
implement
the
remote
of
the
us
execution
and
this
week
or
next
week
we
will
end
other
codes
like
the
same
subs
and
other
things.
B
B
C
What
so,
let's
go
download
next,
this
one,
the
edge
core
as
a
system
server
is
this
is
in
almost
it
should
be
done,
but
I
will
use
your
pr
tomorrow.
B
B
B
So
this
one-
actually,
I
think
it's
I
can
just
quickly
show
you
it's
actually,
but
we
probably
need
a
to
move.
The
location
somewhere
in
the
slab
also
reported
that
it's
it's
hard
to
find
that
it's
hard
to
find
the
location
of
a
different
document,
because
it's
it's
kind
of
located
in
different
directories.
So
actually
like
the
running.
C
Yeah,
because
I'm
also
not
proposing
yeah,
also
I'm
modifying
the
installation
because
I'm
doing
the
vm
reasoning.
So
that's
why
I'm
also
make
change
probably
in
the
wrong
place.
Maybe
I
will
take
a
look.
B
Also,
we
need
add
index
in
the
mentor
too
much.
E
B
C
And
the
the
provisioning
we
are
have
some
issues
because
in
the
documentation
we
recommended
in
the
0.
C
B
C
Yes,
that's
why
it
doesn't
for
the
vm
provisioning
is
not
working
with
zero
that
one
recommended.
I
try.
I'm
testing
0.3.1.
C
B
C
No,
but
we
need
to
install
that
because
after
we
install
vm,
we
need
a
cni
plug-in
to
hook
up
the
yeah.
B
I
think
that
should
be
fine,
because
currently
even
the
installer
don't
help
to
install
that
it's
kind
of
manual
step
for
the
end
users.
So
we
just
need
to
clarify
that
in
the
document.
C
B
C
C
I
think
that's
only
you
can
probably
put
a
memo
into
this
one.
That
means
this
edge
only
support
either
vm
or
container.
We
cannot
mix
it
up
yet
because
if
you
want
mixed
up,
you
need
a
metadata
change,
maybe
from
upstream
so
I.
C
That's
kind
of
user
manual.
C
This
is
a
phase
one
in
the
future
phase,
because
in
the
mec
discussion
we
talked
about
the
mix
it
up.
Also,
I
I
was
in
the
meeting
another
comedy,
meaning
actos,
another,
the
one
phone
was
there,
he
talked
about
the
he
may
need
one
phone
and
a
monkey
were
there.
They
they
talk
about.
They
want
to
mix
up
not
only
in
the
system
but
also
maybe
in
the
same
node.
B
So
I
think
I
would
suggest
that
we
have
a
proposal
to
describe
yeah.
B
C
This
one,
I
think,
we'll
go
with
the.
Let
me
see
this
question.
I
don't
think
the
iot
part
need
the
vm
support
right.
B
To
the
mvc.
C
F
B
Yeah,
so
the
kubernetes
dependency
of
the,
so
actually
this
one
should
be
easy.
We
just
need
to
find
the
time
to
do
it.
C
And
also
test
that
right
we
can,
after
we
call
free.
We
do
that
yeah,
you
can
put
a
say
after
branch
to
verify.
B
B
B
B
So,
basically,
currently
a
set
of
community
members
are
discussing
about
the
the
reference
architecture
and
the
vision
of
mec
and
another
set
of
people
are
discussing
more
about
manufacturing
and
iot
and
the
device.
B
B
So
this
is
a
very
brief
background
about
this,
so
I
just
initialized
the
document
document.
Please
remember,
remember
that
it's
not
a
fi,
it's
not
if
I
know
our
model,
so
I
think
we
currently
currently
for
the
first
phase.
I
think
we
just
need
some
sigs
special
interest
groups
and
when
the
as
the
community
goes
larger
and
larger,
we
may
add
more
types
of
working
groups
or
communities
sort
of
things.
B
So
basically
I
mean
in
general.
The
model
is
very
similar
to
the
kubernetes
model
and
also
we
are
reusing,
some
of
the
well-known
words
with
its
original
meaning.
B
So
I
will
just
quickly
go
through
this
document
so
for
general,
proper
principles
I
think
cube
beige.
We
want
to
build
an
open
source
community
and
embarrass
open
governance,
so
the
things
need
to
highlight
here
is
we.
We
want
all
the
work
to
be
done
in
public
and
transparent
to
especially
new
community
members.
B
So
also
make
all
the
history
accessible
for
for
the
people
joining
later
on
for
the
code
of
conduct
we
have.
We
have
some
clarification
already
in
the
repository.
B
So
basically
cuba
follows
the
cncf
code
of
conduct
and
also
so
so.
I
also
pick
one
paragraph
here
so
basically
just
to
tell
people
we
we
want
to
every
everyone
become
positive
in
the
community
for
our
community
membership.
We
already
have
a
document
there,
so
basically
we
have
different
levels
of
contributor
in
the
community.
B
But
actually
we
have,
we
do
have
reviewers,
so
we
may
need
to
clarify
about
this
and
also
currently
the
the
privileged
discussion
here.
Description
here
is
more
likely
access
to
the
repositories,
but
we
are
actually
relying
more
and
more
on
probot.
B
So
this
the
details
about
the
privileges
we
may
need
to
upgrade
refresh.
B
Anyway,
so
for
the
governance
here,
the
new
new
content,
especially
at
this
time,
is
about
the
description
of
special
interest
groups.
B
B
And
also
regarding
the
cra,
currently
we
don't
we
not
yet
enforce
the
theory.
Checking
it's
not.
A
B
Okay.
I
won't
go
through
details
about
this,
so
I'll
just
send
the
links.
B
So
you
can
take
a
look
of
the
meeting
and
please
give
your
comments
on
the
pull
request.
A
B
So
I
also
I
want
to
highlight
that
the
governance
model
is
more
likely
describe
how
how
the
community
is
organized
currently
and
so,
especially
for
the
sikhs.
What
what
does
a
sikh
look
like,
but
for
studying
a
new
things?
I
think
we
can
it's
a
it's
actually
depending
each
on
each
actual
seat.
So
I
think
for
the.
B
Yeah
or
maybe
we
just,
we
can
just
have
one
first
little-
seek
format
it
and
make
it
as
an
example.
That's
much.
B
B
B
If
no
questions
about
the
governance
model,
we
can
go
on,
so
I
think
these
topics.
B
C
F
All
right
cool,
thank
you.
So
what
I'm
trying
to
do
is
have
machine
learning
operations
on
the
edge,
especially
in
the
academic
environment.
F
F
I
done
cube
h
on
gcp
able
to
put
the
that
installation
well,
but
the
next
stage
was
now
on
deploying
tube
flow,
because
keep
flow
is
already
a
machine
learning
approach
which
tries
to
work
on
kubernetes
and
the
reason
why
I
want
to
use
cube
a
cube,
cube
egg
is
because
there's
already
implementation
of
you
know
master
in
the
cloud
and
then
you
have
the
the
working
notes
on
the
edge.
F
This
will
work
pretty
well
in
the
academic
environment,
to
start
learning
kubernetes,
and
also
to
be
able
to
implement
machine
learning
operations
where
we
only
save
the
model
on
the
edge,
but
we
train
the
models
in
the
cloud.
So
my
question
to
the
community
currently
is:
is
cube
edge.
Is
it
possible
to
run
keep
flow
on
cube
edge,
and
if
that
is
so,
can
I
also
run
cube
edge
on
a
raspberry
pi.
C
I'll
say
you
guys
understand
correctly:
you
want
to
do
the
training
on
the
cloud,
then
you
could
edge
and
push
your
model
down
to
actually
to
run
the
inference.
That's
what
I
understand
is
that
true,
yes,
yes
yeah!
Actually,
on
the
top
of
this
page,
you
can
see
we
have
a
new
blueprint
project
in
chrono.
C
C
Test
rt
because
tess
rt
only
do
only
support
inference,
but
the
only
thing
with
the
limitation
is
pence.
Rt
is
the
developer
by
nvidia.
So
you
need
an
immediate
gpu
to
support
it
to
run
the
model,
but
that's
only
one
example:
we
show
we.
We
could
have
more
examples.
I
we
try.
The
raspberry
pi
before,
however,
is
overloaded.
C
C
However,
if
you
have
a
gpu
the
setup
we
currently
have
for
data
blueprint
project,
it
definitely
can
run
and
that
project
will
provide
an
open
environment
you
can,
after
it
got
approved
by
the
community.
We
are
going
to
open
the
ip
and
everybody
can
come
and
take
a
look
and
they
can
re
produce.
I
mean
reproduce
the
same
environment
in
your
own
environment.
F
Yeah,
it
does
answer
my
question
because
the
thing
is,
but
the
reason
I'm
asking
this
is
before
I
go
deeper.
I
would
look
for
a
different
use
case
because
for
me
I
want
to
learn,
cube
edge
and
I
want
to
use
qbh
to
bring
it
into
the
academic
research
groups.
So
at
first
I
thought
ml
ops
would
be
ideal,
but
I'm
not
sure
if
anyone
in
the
community
has
a
different
use
case.
F
We
could
use
for
for
cube
edge
when
it
comes
to
learning
kubernetes
and
also
understanding
performance
issues
of
orchestration
on
the
edge.
C
Yep,
I
think
that
should
help
you,
I
mean
and
you're
welcome,
to
join
the
meeting.
Where
I
talk
about
this
project
in
the
following
meeting
too,
and
then
we
can
give
you
an
update
and
bring
it's
it's
more
like
an
inference
on
the
edge
or
even
on
the
mobile
device.
It
doesn't
run
tf
lite
instead
of
the
full
tensorflow.
C
F
B
You
all
right,
so
we
have
done
the
agenda.
Anyone
have
any
thing
to.
A
B
So,
by
the
way
regarding
the
annual
review
and
project
intubation,
so
one
step
of
that
is
that
we
we
recently
added
a
doctor's
file
to
the
project.
B
B
B
C
C
So,
as
I
follow
up
in
the
previous
problem,
yeah
so
up
in
the
next
release,
we
may
have
a
roadmap
to
add
more
examples.
I
mean
the
current
example
will
follow
the
lt
ones,
but
we
can
more
examples
after
1.3.
C
B
And
by
the
way
we
we
actually
in
the
early
days
of
the
project
launch,
we
have
some
connection
with
catacoda
and
we
still
have
pending
courses
there.
Maybe
we
can
go
on,
I
mean
after
we
have
more
examples.
We
can
also
launch
corresponding
courses
there.
B
I'm
not
sure
if
we
have
any
existing
example
there.
So,
basically,
I
think
this
is
very
useful
for
the
new
new
users
new
developers
to
try
out
qh
color
code.
It's
kind
of
a
turnkey
environment.
B
B
E
B
C
C
B
Thank
you,
so
the
next
time
would
be
april
29
it's
the
euro
time
next
week.
B
May
may
1st
to
may
5th
so
yeah,
so
so
the
may
6th.
We
are
back.
So
it's
the
it's!
The
the
next
next
meeting.