►
From YouTube: Antrea Community Meeting 09/12/2022
Description
Antrea Community Meeting, September 12th 2022
A
Good
morning
or
good
afternoon
and
welcome
to
the
andrea
community
meeting
today
is
tuesday
september
15th
and
for
today
we
don't
have
any
items
so
far
in
the
agenda,
so
it
will
be
pretty
much
open
discussion.
A
Okay,
I
believe
that
we
can
get
started
not
very
wide
audience
for
today.
So,
as
we
said,
there
is
no
topic
on
on
the
agenda,
so
if
anybody
from
the
attendees
would
like
to
propose
any
topic
to
discuss,
please
please
go
ahead.
A
All
right
so
doesn't
seem
that
we
have
any
topic,
so
I
would
like
just
to
spend
a
couple
of
minutes
providing
a
quick
update
on
the
on
the
status
activities
for
the
tia
release.
As
you
know,
you
know,
ti
is
the
network
observability
project
we
are
in
the
process
of.
A
We
are
in
the
process
of
introducing
a
new
component,
which
we
are
calling
conveniently
calling
the
tia
manager,
and
the
task
for
this
component
will
be
to
provide
a
an
interface
for
interacting
with
the
network
policy
recommendation
engine,
at
least
for
this
first
release.
Then
this
will
be
sort
of
become
the
let's
say,
the
entry
point
for
controlling
all
the
andrea
observability
features.
A
In
addition
to
that.
So
so
far
we
only
have
merged
a
first
pr
which
creates
sort
of
the
boilerplate
for
the
server,
and
we
are
in
the
process
of
then
adding
the
adding
the
apis
that
will
process
the
various
crs
I
mean,
and
for
the
cities.
For
this
first
release
we
pretty
much
have
just
a
network
policy
recommendation
cr.
A
We
did
not
present
it
in
this
meeting
because
we
still
don't
have
let's
say
something
that
works
end-to-end,
which
can
be
which
can
grant
a
nice
demo.
That
will
be
something
that
we
will
probably
discuss
in
the
next
community
meeting.
A
Apart
from
that
for
the
tier
release
you,
we
will
have
a
new
graphene
page
that
was
already
presented
a
couple
of
meetings
ago
and
yeah,
and
that
would
be
pretty
much
it
in
terms
of
important
features
that
will
be
added
in
the
next
tier
release.
I
don't
know
if
anyone
from
yet
indeed
wanted
to
share
a
little
bit
of
the
progress
related
to
the
features
that
are
going
to
be
emerging,
andrea
1.9.
A
All
right,
so
there
doesn't
seem
to
be
much
to
be
updated
regarding
the
release
status
for
for
this
week.
And
yes,
I
just
wonder
if
you
have
anything
that
you
would
like
to
bring
up
in
terms
of
general
discussion,
process,
improvement
or
you
know,
unit,
test
coverage
or
any
other
topic
which
might
be
important
to
discuss
with
the
community.
If
so,
please
go
ahead.
B
A
C
Challenge
we
have
a
question
on
united's
college
gating
and
we
added
the
gate
for
pr
merge.
Saying
the
new
chain
should
have
should
reach
70
percent
coverage.
C
I
just
want
to
know:
what's
the
current
situation
for
that
gating,
do
we
make
it
monetary
already
or
the
thing
for
particular
cases,
it
can
be
exception.
B
It
is
made
mandatory,
but
we're
merging
some
prs.
It
is
really
hard
to
make
it
achieve
that.
So,
while
a
few
of
them
skipped
the
I
I
first,
I
first
skipped
the
check
because,
for
example,
if
you
want
to
remove
some
stair
coat,
I
think
that's
accurate
as
a
thief
but
yeah
you,
don't
you
there's
no
point
to
add
any
test
cut
a
new
test
code
for
that
remote
code.
C
Because
I
I
did
say
a
few
pr's
didn't
reach
70
coverage
and
for
most
of
them
I
asked
the
owner
to
try
to
increase
the
coverage
rate
yeah,
but
I
just
wanted
to
have
any
generic
principle
about
such
situations.
Should
we
generally
ask
the
author
to
try
to
improve
the
reach
coverage
target
or
in
many
cases
we
can
have
exceptions.
A
A
Yeah
I
mean
I
just
have
a
question
for
the
for
t:
I'm
thinking
that
the
unit
test
constraints
coverage
constraints
will
need
to
be
enforced
on
python
and
golang,
because
basically,
let's
say
that
it's
the
code
base
it's
half
enough
between
python
and
golem
at
the
moment,
because
you
know,
as
you
know,
for
network
policy
recommendation,
everything
is
written
in
python.
A
I
wonder
if
I
mean
if
we
have
a
pr
which,
where
we
only
have
a
golan
code
or
only
have
python
code,
it
should
be
easy
to
enforce
a
python
job
that
does
that
does
a
unit
test
coverage
either
for
python
or
for
googling.
Then
I
wonder
if
we
have
any
pr
which
changes
both
python
and
golden
code,
whether
we
need
something
special
to
enforce
a
getting
criteria
on
having
minimum
level
of
coverage
both
for
the
python
code
and
the
cooling
code.
A
I
I
don't
think
it's
impossible.
I
just
wonder
if
you
believe
there
might
be
some
difficult
difficulty
in
achieving
that
with
code
cove.
A
Yeah,
that's
that's
the
same.
I
believe
that
what
we
need
the
property
to
do
is
that
we
need
to
run
separately
the
job
jobs
for
python
and
googling,
and
then
we
should
have
some
other
some
other
task
that
compares
that
both
results
are
x
are
be
above
a
certain
threshold,
but
anyway,
yes,
it's
just
a
technical
thing.
We
need,
we
need
to
verify
whether
it's
possible
or
not,
okay
and
right.
A
Okay,
so
waiting
a
few
more
seconds
to
wait.
If
there
is
anybody
from
the
team
with
any
other
topic
for
discussion
for
today,.
A
B
You
under
10
in
one
time
yeah
we
only
plan
to
finalize
the
design
and
poc
okay,
and
since
we're
already
talking
about
this,
I
have
a
question
about
license.
I
checked
the
tours
like
ricardo
his
license.
This
license
is
gpl
version
two
and
I
checked
that
it
is
similar
to
other
linux
tools
like
tables
and
net
filters.
So
I
assume
it
is
okay
to
package
the
upstream
build
binaries
in
and
just
deliver
levels
and
use
that
tool
directory
in
android,
and
am
I
right?
A
Yes,
with
the
gpl
v2,
usually
re-packaging
does
not
pawn
any,
does
not
there's
no
licensing
challenge
with
the
repackaging.
A
It's
slightly
different
in
terms
of
attribution
when
you
do
modifications
to
the
original
code,
but
yeah
there
should
be
no
problem.
If
it's,
if
it's
a
gpl
v2
license,
then
we
are
just
re-packaging
the
binaries,
and
even
if,
even
if
we
were
since
at
the
end
of
the
day,
andrea,
it's
still,
it's
still
all
fully
open
source.
It
would
not
have
been.
A
problem
is
just
that
the
overall
andrea
license
would
have
changed
as
well
becoming
apache
plus
gpl
v2.
A
But
since
we
are
just
re-packaging
the
package,
there
should
not
be
any
problem,
but
in
any
case
yeah,
that's
something
that
I
can
verify.
A
I
did
some
licensing
stuff
in
the
past,
so
I
remember
a
little
bit
of
this
thing
so
I'll
check.
What
are
the
implications
on
the
of
repackaging,
gpl
v2
components,
because
you
know
the
difference
with
ip
tables
is
that
it
is
through
the
type
enables
this
gpl
v2,
but
we
are
not
redistributing
it
in
this
case.
I
believe
that
we
are
redistributing
this
uricata
and
that's
slightly
different
from
my
variables
from
the
way
we
are
using
epitables.
B
A
Yeah
yeah:
that's
that's
a
good
point
that
stunt
amount
to
distributing
it.
Yes,
you're
right,
that's
a
good
point.
So
yeah.
C
We're
using
many
gpl
lessons
components
right,
for
example,
open
with
switch.
C
A
A
Even
just
by
building
a
ubuntu
based
image
about
from
my
pd
box,
we
are
completely
redistributing
many
gpl
v2
components.
So
it's
I.
I
don't
think
it's
a
problem,
but
it's
a
good
point.
It's
been
raised
by
chan
and
it's
better
to
verify
whether
we
also
need
to
change
some
attribution
in
the
entry
a
license
because
of
this
so
yeah.
That's
something
worth
verifying
should
not
be
a
problem.
However,.
B
Yeah
I
raised
the
question
because
I
saw
that
the
license
way
accept.
We
accept
it.
We
set
we
when
we
configure
the
license
checking
tool.
Gprwa2
is
not
in
the
list,
but
I'm
not
sure
whether
it
is
because
we
that
tool
is
only
used
to
check
the
code,
not
the
binary.
B
A
A
Yeah
and
chan,
one
more
question
for
me
in
terms
of
discussing
and
presenting
and
seeing
a
demo
of
layer,
seven
policy:
do
you
think
that
we
can
do
it
like
in
the
next
meeting
in
a
couple
of
community
meetings.
B
Yeah,
you
you
mean,
have
a
presentation
about
the
design
and
poc
yeah.
B
Yeah
yeah,
I
I
think
so,
okay,
but
no,
no,
I'm
not
sure
whether
next
community,
maybe
I
could
try
yeah.
C
A
Ready-
and
this
is
a
target
for
andrea
one
pointer-
and
you
say
it
right:
yes,.
A
Yeah,
that
would
be
great,
and
and
just
in
terms
of
the
potential
design
are
we
introducing
new
custom
resources
or
just
modifying
the
existing
custom
resources
for
entry
network
policies.
B
I
think
we
may
need
to
introduce
a
new
one
because
it
cannot
be
enforced
with
us
while
the
same
priority
mechanism,
because
it
cannot
you
you
cannot
insert
or
and
append
lures
before
and
after
the
seminar
policies.
It
must
be
like
kind
of
for
kubernetes,
narrow
policies,
no
priority
and
no
tail
just
a
black
list
or
a
white
list
so
understood.
It
cannot
be
easily
combined
with
internal
policies
with
current
gravity
and
the
tl
design.
A
Understood
perfect
thanks
a
lot,
so
we
look
forward
for
reviewing
the
design
and
and
having
a
nice
theme
of
this
feature
as
soon
as
possible,
and
is
there
any
other
question
for
chan
or
any
other
topic
that
you'd
like
to
bring
up
for
today.
A
And
therefore,
I
believe
that
might
be
all
for
today,
so
thanks
everyone
for
joining,
I
wish
everyone
a
great
day
a
great
afternoon
or
good
evening
or
a
good
night.
So
thanks
again
for
joining,
and
I
will
meet
again
in
two
weeks
time
and
if
you
have
anything
that
you
would
like
to
present
anything
that
you
would
like
to
discuss
anything.
That's
bothering
you
please
let
us
know
in
the
andrea,
the
luck
channel
slack
channel
so
that
we
can
schedule
your
topic
in
the
next
agenda.