►
From YouTube: Nebula Graph community meeting [2022-04-06] (Balance Data is back, Nebula Graph on Raspberry Pi)
Description
🚩 Project heartbeats
- 3.0.2 released with ARM support
- #3997 support `MATCH p=(v:player)-[:serve]-() WHERE (v)-[:follow]-() RETURN p` now
- #3929 support `MATCH (v)--(v)` now
- #3677 More comments in codebase
- #4085, #4067 more information on `SHOW JOB `
- #4088, #4017 Balance Data
- #4100 Notice in Storaged service status for Not added hosts
- Nebula K8s Operator Supports 3.0 Now
Ad-hoc Topic:
Nebula Graph on SBC/Raspberry Pi
- https://www.siwei.io/en/nebula-graph-on-pi/
A
A
A
A
A
A
So
the
major
another
thing
that
I
didn't
put
in
this
slice
is
we
we
have.
We
will
have
the
3.0
release
in
upcoming
weeks
and
the
first
release
candidate.
The
first
rc
is
also
tagged
in
a
branch
called
release,
heaven
3.1
and
you,
if
you're
interested
in
that
you
can
check
that
release
as
well.
But
it's
the
major
new
contents
comes
from
the
master
anyway.
So
so
in
last
past
two
weeks
in
master
branch,
we
have
a
couple
of
major
prs
being
merged.
A
The
very
first
one
is
that
we
finally
support
this
type
of
query
that
you
can
put
a
pattern
in
the
where
clause,
so
this
is
very
handy
for
certain
use
cases
and
is
being
required
for
some
community
users
so
boom.
We
finally
have
that.
Another
thing
is
that
previously
we
we
have
some
issues
when
in
a
pattern,
one
of
the
allies
or
eyelids
or
or
or
of
very
variable
being
claimed
twice,
so
it
it
will
not.
A
It
was
not
supported,
and
now
it's
supported
here-
and
this
is
a
huge
issue
that
comes
with
like
10
of
pr's
as
a
subtest,
that,
with
the
jointly
efforts
from
the
whole
community
and
our
contributor
has
made
it
work
that
we
have
so
many
more
comments
on
the
code
base.
So
to
me
that
when
I
was
trying
to
make
some
contributions
on
the
graph
engine
that
I
know,
I
noticed
that
a
lot
of
the
code
base
is
not
easy
to
be
understand,
especially
for
the
new
commerce
on
the
database
area.
A
And
now
it's
it's
much
better
that,
with
the
works
of
the
last
few
months,
a
lot
of
new
comments,
even
with
some
ask
art
diagrams
being
introduced
in
in
the
comments.
So
it's
it's
it's
much
better,
but
we
are
making
it
even
better
in
in
in
upcoming
months
as
well,
but
this
is
a
huge
milestone
so
be
sure
to
check
out
the
code
base.
There
are
a
lot
of
beautiful
comments
now,
and
another
thing
is
that
in
a
show
show
job
we
have
made
some
made
some
improvements
that
the
the
field
jobs.
A
A
Another
thing
is
the
balanced
data,
as
we
know
from
the
3.0.
For
some
reason,
we
noticed
some
unstable
ability,
stable
reasons.
There
will
be
some
rare
case
that
balanced
data
will
failed
and
will
cause
some
data
loss.
So
we
put
the
balance
data
function
on
hold
from
the
3.0,
and
now
we
add
it
as
a
experimental
feature
back
to
the
master
and
from
3.1
the
balanced
data
will
be
usable
again
as
an
experimental
feature.
A
We
implemented
that
in
a
different
way
in
other
enterprise
version,
because
we
have
the
zone
support
in
enterprise
version.
So
this
is
a
big
change
for
the
user,
so
3.0
is
quite
usable
as
a
production
candidate,
as
I
can
consider.
Another
thing
is
that
storage
service
we
we
have
known
that
from
the
3.0.
A
We
implement
a
new
mechanism
on
the
storage
service
discovery
because
previously,
if
the
service
is
off
and
it
notified
the
meta
service
that
it
is
off,
then
it
can
be
used.
It
is
marked
as
the
online
in
the
service
states,
but
now
we
have
the
actual
step
before
it's
being
online.
We
market
it
registered
as
offline,
but
it's
it
needs
news.
A
The
user,
the
manual
inventions
to
explicit
ad
hosts
to
actually
active,
activate
the
service
to
be
marked
as
online
and
usable,
but
this
brings
confusion
to
the
users
that
was
ex
experienced
with
the
2.0
2.x
version
of
nabla
graph,
so
they
will
only
see
the
offline
in
the
service
status
command.
So
we
add
actual
notificat
highlights
here
that
you
need
to
do
the
add
host
to
make
it
online
instead
of
confusing
suspension
on
the
service
states
itself.
A
A
A
So
I'm
I
try
this
in
in
my
sbc,
which
is
not
a
reciprocal
pipe
or
similar.
So
this
is
the
pie
I
got.
It's
called
rox
pie,
it's
a
similar
sbc,
but
with
the
neural
precise
unit
mpu.
So
maybe
I
will
leverage
in
this
mpu
capability
to
running
the
you
know
the
machine
learning
workloads.
So
maybe
I
can
create
something
to
share
with
you
in
upcoming
months.
A
I'm
thinking
about
that.
So
you
can
run
network
memograph
on
on
this
kind
of
board,
because
nabla
graph
comes
with
support
of
both
intel
architecture,
the
arm
and
even
mips,
and
as
I
mentioned
in
from
next
version,
the
nav
graph
will
come
with
the
docker
image
out
of
box
supporting
arm-based
system.
A
Actually,
the
the
best
way
running
a
naval
graph
on
a
single
on
the
edge
case
on
the
small
board,
as
you
are
running
the
single
board
field.
So
we
introduced
this
build
option
things
3.0,
so
we
removed
the
rpc
call
between
the
the
graph
and
the
storage
service.
We
combine
them
in
in
the
one
process,
so
they
will
become
smaller
and
thinner
overhead.
A
So
it's
actually
more
ideal.
Smaller
food
footprint,
when
you
are
actually
running
this
on
pro
production
in
the
ash
computing
case
smaller
as
this
spc,
but
but
in
this
sharing,
I'm
just
to
prove
it's
a
park
that
is
working
so
I'm
using
a
easier
way,
but
it's
not
the
best
way,
I'm
using
it
actually
on
with
the
nebular
graph
doc
compose
so
underlying
is
still
much
more
processes.
A
It's
a
cluster
still,
but
you
don't
need
the
catharine
in
in
the
pipe
right,
but
we
can't
yeah
and
in
this
article
on
my
blog,
I'm
leveraging
the
docker
compose,
and
this
is
the
board.
It's
awesome
right,
and
so
what
I
do
is
I
I
bootstrap
ubuntu
on
this
card
single
spc
and
I
installed
docker
and
then
I
just
installed
docker
compose-
and
this
is
a
blocking
issue-
that
the
official
documentation
on
docker
to
install
docker
compose
is
not
valid
for
the
sbc,
because
it's
not
supporting
the
arm
properly.
A
So
I
will
demo
it
later
and
you
can
even
have
this
arm-based
console,
which
is
the
client
command-line
client
for
the
number
graph
it.
It
comes
with
the
both
windows
mac
arm
and
the
an
intel
linux
support.
So
after
you
give
a
permission
of
execution,
you
can
run
this
on
your
local
support,
local
host
and
be
sure
to
add
host.
As
I
mentioned,
this
is
needed,
or
your
storage
service
will
not
be
opened
online.
A
Then
you
can
play
with
your
console
so
use
the
playback
player.
If
this
test,
you
know,
is
the
same
data
data
set
in
our
documentation
will
be
imported
to
the
nanograph.
It
will
take
you
around
one
minute
and
then
you
can
use
this
from
the
console,
use
the
the
graph
space
and
query
the
ngql,
and
now
I
will
give
you
a
small
demo
on
that.
We
will
do
it
from
scratch.
A
A
and
I'm
using
the
cluster
version,
and
you
can
see
everything
is
up
and
running
in
different
replica,
and
we
can
we
I'm
using
this
client.
Also
from
this
pie,
it's
very
cute
and
we
are
using
this-
they
have
base
space
and
I'm
giving
the
same
query
that
I
introduced
in
the
project
updates
that
we're
matching
all
the
player
that
serves
something
player
serves
something
serve
and
where
the
the
version
are
with.
A
You
know
following
another
player,
so
this
is
a
match
pattern
in
the
weird
clause,
so
oops,
it's
quite
responsible
right,
responsive
right
and
we
can
see
it's
not
that's
consuming
that's
memory
consuming
and
we
can
see
it's
quite
usable
already
right.
It's
not
consuming
a
lot
of
resources,
so
it's
quite
cool
right.
I'm
I'm
looking
forward
to
make
something
even
cooler
on
on
this
board
with
nabla
graph
later,
and
I
hope
I
will
share
with
you
around
that
too.
A
So,
okay-
and
that
is
all
around
this
ad
hot
topics.
Then
we
will
move
on
to
next
part.
Okay,
we
already
demo
that
okay,
so
we
are
making.
I
will
call
this
an
end
before
that.
So
remember
we
are
having
this
by
weekly
and
you
will
have
be
sure
to
subscribe.
Our
our
meeting
in
the
zoom.
We
have
a
google
calendar,
it's
a
shared
calendar
and
everything
here
it
will
be,
can
be
fetched
from
our
wiki
and
be
sure
to
join
our
stack
channel
and
once
and
before
we
call
this
end
this.