►
From YouTube: Octant Community Meeting - June 16th, 2021
Description
Octant community meeting is held weekly. We discuss and talk about the current state and future of Octant, demo upcoming features and releases, and preview new ideas we are considering for Octant.
Meeting agenda: https://hackmd.io/CzaPxtmXT_SW8nEpdwvGzw?view
A
B
A
You
know-
probably
not-
I
can't
guarantee
any
of
this,
but
you
know
we
are
closer
in
the
sense
that
they
are
button
clicks
away
as
opposed
to
movements
away,
but
anyway
we
will.
We
promise.
We
are
trying
very
hard,
we're
gonna,
think
all.
A
E
Oh
yeah,
I
can
talk
about
it
sure,
okay!
Well,
basically,
we've
been
looking
into
how
kind
of
set
a
single
responsibility
to
the
dash
dot
go
that
should
go
is
very
much
used
to
initialize
a
lot
of
services
and
other
yeah
different
things,
and
the
main
idea
was
to
be
able
to
separate
this
logic
from
our
main
function.
E
E
The
main
idea
was
to
be
able
to
have
a
plan
where
we
can
have
a
group
of
concepts
and
a
deployment
deployment
path,
so
we
can
see
if
we
can
do
it
only
in
once
or
we
can
go,
create
user
stories
related
with
this
and
go
and
clean
it
up
a
little
bit
as
we
go.
So
yes,
I
think
we
I
I
like
it
a
lot
wire
and
it's
something
that
we
should
all
check
out
and
see.
If
which
is
gonna,
do
an
improvement.
A
Awesome
and
thank
you
for
bringing
this
up
explicitly,
because
the
strangeness
is
definitely
something
we've
been
aware
of
even
before
you
joined
this
team
and
we've
never
really
gotten
around
to
addressing
it
head-on,
and
this
is
especially
rampant
even
from
the
testing
level,
all
the
way
to
just
the
amount
of
like
one
we're
seeing
a
ton
of
race
conditions
in
the
code
and
also
we've
got.
We've
see,
we've
gotten
a
lot
of
weird
patterns
as
a
result,
and
we
just
keep
talking
about
dash.go
and
really.
A
This
is
this
is
inherently
the
problem
right,
and
so
we
just
need
a
better
way
to
model
how
how
we're
starting
up
dependencies
in
octane
and
also
their
entire
life
cycle
as
octane
is
running.
B
Cool
yeah.
No,
this
is
really
great.
I'm
excited
to
see
where
this
goes.
It's
been
something
like
sam
said
that
we've
had
on
our
radar
for
a
long
time
and
now,
with
the
the
team's
size
being
increased,
we
actually
can
start
to
attack
things
like
this,
which
is
pretty
great.
I'm
yeah
super
excited
for
this,
because,
thanks
again,
louise.
F
A
E
Yes,
we
we
can
set
up,
like
you
know
how
often
data
flow
works
in
a
diagram.
This
is
something
that
there
is
a
this
document
is
a
very
high
overview,
doesn't
have
too
much
detail
and
it's
very
short,
the
idea
will
be
to
build
on
it
and
try
to
put
different
things
like
the
recommendations
you
created
some
together
with
diagrams
that
bring,
has
and
I'll
be
able
to
collect
all
this
information
in
one
place.
E
So
if
somebody
wants
to
contribute,
it
will
be
an
easy
way
to
start
know
how
the
data
flow
works
in
often
and
and
yeah.
This
yeah
that's
pretty
much.
E
I
don't
know
if
you
guys
have
time
to
to
check
the
document,
but
as
soon
as
we
have
something
a
little
bit
more
robust
robust,
it
would
be
nice
to.
B
B
That
we
create
a
github
discussion
to
start
tracking
the
like
what
we're
capturing
here
and
the
conversations
we've
had
and
get
them
in
a
place.
That's
a
little
more
long-lived,
and
then
I
don't
want
to
worry
too
much
about
like
it
being
in
draft.
I
would
rather
just
go
ahead
and
and
share
what
we
have
now
and
we
can
start
building
on
it
out
in
the
open.
Like
it's.
It's
light,
that's
fine!
B
A
All
right
sounds
good.
We
will
be
ready
to
get
up
discussion
and
well
yeah
and
luis
if
you
want
just
even
a
first
pass.
So
just
for
thing
like
for
what
you
have
so
far,
even
just
for
like
sanity,
checks
or
simple,
like
simple
read,
like
reading
type
checks,
I
can
definitely
take
a
look.
B
Yes,
I
put
this
one
up
here.
This
was
something
that
I
started
looking
at
a
couple
weeks
ago,
but
then
I
was
out
for
a
bit,
and
so
I
got
a
chance
to
circle
back
to
it.
A
little
more
recently
I
didn't
have
the
demo
that
I
was
hoping
to
have
prepped,
because
I
had
I
got
pulled
off
to
look
at
some
of
the
streaming
pr
work
that
I
wanted
to
make
sure
got,
got
landed
pretty
soon,
since
we
have
another
team,
that's
depending
on
it.
B
So,
but
the
the
short
of
this
is
is
that
kcp
is
right
there,
a
minimal
control
plane
for
cube
like
applications.
It
creates
a
kubernetes
api
server,
it's
essentially
empty,
and
then
you
can
fill
it
with
your
resources
and
then
query
it
like
it's
a
kubernetes
api.
One
of
the
things
that
this
tool
has
is,
it
has
a
feature
called
the
syncer.
The
syncer
is
essentially
very
similar
to
what
octant
does
now,
with
its
object,
store
and
its
dynamic
cache.
It
goes
to
a
cluster
fetches.
B
B
So
the
interesting
thing
about
this
is
that
octant
itself.
You
know
it
implements
a
lot
of
this
logic
and
it's
trying
to
do
this
work,
but
kcp
being
a
project
dedicated
to
this
purpose
means
that
it's
going
to
actually
have
a
better
like
it
can.
B
It
has
to
worry
about
less
problems,
and
so
it
can
really
focus
on
doing
this
thing
well,
which
is
syncing
resources
from
a
cluster
or
multiple
clusters.
So
I'm
I'm.
B
I've
spoke
to
that
team
and
and
their
slack
channel
they're
open
to
the
idea
of
of
well
they're,
really
excited
about
running
on
top
of
kcp
in
general,
just
because
I
think
we'll
both
get
some
good
feedback
loops
out
of
it
and
be
able
to
improve
each
project,
but
they're
also
open
to
the
idea
of
accepting
pull
requests
that
would
enable
octane
to
kind
of
leverage
kcp
as
a
place
to
store
its
cube,
config
information
so
that
underpinning
octant
is
now
a
instance
of
kcp
running
versus
what
we
have
now,
which
is
our
own
custom
object
store.
B
The
another
thing
that's
just
an
interesting
detail
about
this
is
that
if
you
go
look
at
the
kcp
code,
the
syncer
code,
that
they
have
it's
like
it,
it's
almost
identical
to
what
we're
doing
in
our
object
store
right.
It's
like
sam-
and
I
were
talking
about
this
a
little
bit
about
how
like
it's
really
like
three
ways
to
get
objects
from
a
kubernetes
cluster.
There's
like
you
can
use
the
watch
api
like
standard.
You
know,
grpc,
watch
api
and
pull
things
down.
B
You
can
create
a
shared
informer
right,
and
this
is
all
for
out
of
cluster
in
cluster
is
a
little
different,
but
out
of
cluster,
you
use
the
watch
api.
You
create
a
shared
informer
at
clusterscope
and
you
get
everything
or
you
create
namespace
scoped
informers
and
you
get
things
for
a
particular
namespace.
B
These
are
like
pretty
much
the
three
standard
ways
to
get
this
information,
and
so,
when
you
start
to
go
and
look
at
all
the
all
the
code,
that's
out
there
for
various
projects
that
are
doing
a
thing,
which
is
fetch
resources
from
kubernetes.
It's
really
funny.
You
see
all
the
very
similar
patterns
evolve
and
all
of
the
like
things
that
we've
done
and
had
to
work
around,
and
you
see
those
being
done
in
other
places
and
and
so
for
me.
B
The
motivation
here
is
like
if,
if
kcp
is
able
to
handle
multiple
cube,
config
files
at
once,
it's
doing
a
very
similar
thing
under
the
hood
for
fetching
resources,
then
kind
of
leveraging
that,
as
as
part
of
like
making
it
a
dependency
for
octant,
saying,
hey,
octan's
object,
store,
is
essentially
kcp.
B
Now
I
think
we
we,
if
it
works
out
the
way
I'm
hoping
and
the
project
continues
to
evolve,
the
way
I'm
hoping
we
get
a
pretty
big
benefit
of
it
and
that
benefit
being
a
group
of
people
focused
on
this
concept
of
how
do
we
most
efficiently
effectively
and
reliably
fetch
resources
from
a
cluster
right,
and
it
takes
away
a
whole
class
of
concerns
for
us
as
an
octan
project,
and
we
can
really
focus
on
okay.
B
We
have
all
these
resources,
let's,
let's
really
focus
on
making
them
useful
for
a
developer,
capturing
the
object,
statuses,
making
actionable
resolutions
through
the
ui
things
that
we
really
want
to
get
towards
with
octet
in
the
future.
So
that's
I
just
wanted
to
bring
that
up.
B
I'll,
create
I'll,
create
a
discussion
for
this
as
well,
because
I
think
the
that
way
I
can
link
to
any
upstream
prs
in
kcp,
and
we
can
talk
about
it
as
a
team
and
kind
of
prep
for
any
like
just
like,
exploring
or
demo
stuff
that
I
might
be
doing
there
honestly.
I
don't
expect
it
to
take
very
much
time
to
get
a
demo
up
and
running.
I
just
I
just
need
to
find
a
nice
two-hour
block
to
sit
down
and
do
it.
So
that's
where
it's
at.
B
Yeah
exactly
and
that's
kind
of
one
of
the
things
in
octane
we've
we've
avoided
being
multi-cluster
because
we
didn't
want
to
deal
with
being
multi-cluster.
We
just
didn't
have
the
resources,
whereas
kcp,
as
I
said,
like
they're,
focusing
on
this
thing,
I
think
that's
their
that's
their
domain,
so
we
kind
of
inherit
the
ability
to
be
multi-cluster
from
them.
Potentially,
that
is
that's
just
like
taking
a
guess
in
theory
that
would
work,
but
in
practice
it
might,
it
might
be
a
little
more
challenging
we'd
have
to
see.
B
I
think
it
would
be
more.
I
think
octant
might
might
break
in
weird
ways
or
unexpected
ways,
but
it
would
be
something
we
could
address
to
add
a
little
bit
more
detail
there.
The
way
kcp
does
multi-cluster
now
is
it
produces.
B
There's
a
x
kubernetes
cluster
header
that
gets
embedded
into
the
the
responses
and
the
requests
that
that
tells
you
what
cluster
it's
for,
and
I
believe
they
also
expose
the
forward
slash
clusters
forward,
slash
cluster
name
endpoint
that
lets
you
list
the
clusters,
so
you
kind
of
have
to
at
least
just
my
understanding.
I
think
you
have
to
take
both
of
those
informations
and
and
kind
of
use
them
to
say
here's
my
list
of
clusters
and
also
here's
my
here's.
B
D
Yeah,
I
think
this
is
really
cool
and
I'm
really
excited
about
this.
I
think
especially
because
this
is
probably
the
most
critical
part
of
our
code.
In
my
opinion,
you
know
just
interfacing
kubernetes.
We
had
a
lot
of
issues
there
and
we
you
know
it's
it.
It
was
really
hard
to
find
the
optimal
way
to
collect
all
that
information.
So
if
he,
if
he
can,
you
know,
have
external
layer
doing
that.
For
us,
that
would
be
awesome.
One
thing
on
on
a
negative
side
that
kind
of
throwed
me
off
is
first
thing.
D
I
noticed
when
I
went
to
their
a
kit
is
they're,
saying
it's
a
it's,
not
a
project,
it's
a
prototype,
so
I
think
we
need
to
be
a
little
cautious.
There
I
mean.
Hopefully
you
know
they
they'll
get
to
the
level.
I
don't
know
how
old
the
project
is.
Maybe
then
you
can
talk
more
about
that
and
you
know
what's
the
plan
kind
of
to
maybe
make
it
production
ready.
B
Yeah,
no,
I
think
that's
a
good
point
one.
So
one
of
the
first
things
I
did
was
just
go
and
review
the
code
and
that's
kind
of
why
I
was
calling
out
the
sinker
code
they
have
is.
The
sync
is
like
the
code.
We
have
right
so
I'm
like
well,
even
if
they
abandoned
this
project.
Well,
it's
the
same.
It's
like
it's
the
same,
so
it
doesn't.
B
It's
not
like
it's,
not
a
big
concern
of
mine,
because
I
think
some
of
the
things
they
have
is
already
like
a
good
foundation
for
doing
it
better.
But
the
other
part
of
this
is
that
their
the
team
itself
is
very
motivated
to
to
make
that
like
they
have,
they
have
a
plan,
they
have
a
long-term
plan.
You
know
it's
a
lot
of
the
folks
from
the
open
shift
group
at
red.
B
On
this-
and
you
know
it's-
it's
also
an
apache
2
licensed
project,
just
like
octane,
so
our
usage
of
it
and
taking
advantage
of
what
they
have
now.
Overall,
I
would
say
my
concern
from
that
angle
is
pretty
low,
like
I
think
the
like
they're,
calling
it
a
prototype
just
because
a
lot
of
the
pieces
that
are
multi-cluster
are
very
exploratory.
B
There's
still
like
name
collisions
with
certain
crds,
and
things
like
that
happening.
But
overall,
the
single
cluster
aspect
of
this
is
very
solid
and
and
probably
better
than
what
a
little
bit
better
than
what
we
do
now
in
octan.
Just
because
of
its
ability
to
have
those
individual,
crds
or
individual
clusters
captured
as
crds
with
a
cube
config.
It
will
make
kind
of
managing
them
and
and
showing
a
list
of
them
and
and
importing
and
exporting
them
quite
a
bit
easier
than
what
we
could
potentially
do
now
with
octane.
D
Yeah,
that's
awesome.
I'm
excited
to
see
where
this
is
heading.
It's
almost
like
you
know
having
this
big
part
of
our
code
outsourcing
to
to
another
team.
So
it's
almost
like
a
joining
our
team
right
yeah
we
can,
if
we
can
kind
of
find
a
way
to
work
together.
Even
if
we
have
to
you
know,
do
some
pr's
on
their
side
to
make
it
work.
It's
still
going
to
be
huge
benefit
long
term,
hopefully
for
us.
So
this
is
awesome.
F
Hey
so
I'm
jason,
I
was
randomly
browsing
youtube
and
youtube
said
that
the
octant
community
meeting
was
happening.
I'm
I
work
on
kcp,
I'm
one
of
the
people
working
on
kcp.
This
is
super
serendipitous
yeah.
I
am
more
than
happy
to
answer
any
kcp
questions
you
have
here
or
elsewhere.
I'm
more
than
happy
to
talk
about.
F
You
know,
improvements,
changes
you'd
like
to
make
kcp
here
now
elsewhere.
I'm
I'm
really
excited
about.
The
possibility
of
I
feel
like
octane
is:
is
this
kubernetes
focus
but
mostly
kubernetes?
Api
focus
front
end
for
things
and
it
doesn't
necessarily
need
to
care
about
pods
or
nodes,
or
anything
like
that
and
kcp.
Is
this
very
generic
api
server
idea
and
I
definitely
see
a
lot
of
them
to
use
a
pm
term
synergy
between
kcp?
F
Is
this
generic
back-end
api
and
octane
as
a
generic
front-end
ui
thing
so
yeah?
I
don't
have
like
anything
else
to
say,
except
if
you
have
more
questions
right
now,
I'm
happy
to
answer
them
or
otherwise,
offline
anytime.
This
was
the
most
amazing
surprise
today
to
to
see
that
you
were
talking
about
this.
B
That's
yeah,
that's
great
jason!
Thank
you
yeah!
I
don't
have
anything
outstanding
right
now.
You
already
saw
the
one
thing
I
brought
up
in
the
slack
channel
and
responded
there.
So
yeah,
I'm
not
sure
when
you
at
what
point
you
joined.
I
was
just
kind
of
kind
of
talking
about.
Maybe.
B
F
A
A
D
This
question
yeah,
it's
a
it's,
not
a
trick
question
it's
something
I
run
into
just
when
I
was
working
on
the
resources
and
you
know.
D
Actually,
more
on
a
resource,
viewer's
side
than
expanding
and
collapsing
things
like
you
know,
deployments
and
pods
and
yeah
so
trying
just
to
make
sure
that
we
do
that
properly,
I
mean
there's
always
my
ebay
attending
yes
at
the
end,
which
I'm
doing
now
and
definitely
not
the
way
to
do
it
so
in
kubernetes
world
is
there?
Is
there
like
a
api
or
what's
the
proper
way
to
do.
B
That
yeah,
so
I
mean
there
actually
is
a
pluralize
like
there's
some
code.
I
think
it's
in
util.
B
B
Does
it
that
kind
of
stuff
and
we'll
attempt
to
give
you
back
the
plural
name,
it's
part
of
what's
used
when
you
try
to
convert
like
gvk's
to
gvrs
and
you
do
resource
lookups
that
way
it
runs
through
some
pluralizing
machinery
I'll,
find
it
and
and
be
sure
to
link
it
in
the
notes.
I
don't.
If
anyone
knows
it
offhand,
then
that's
great,
but
I
I
yeah.
I
have
to
dig
for
it.
A
Yeah
to
help
with
that
taking
process,
I
believe
it's
part
of
the
rest
mapper,
that's
the
place
to
start
off
with
so
I'd,
probably
just
look
up
the
rest
mapper
code
and
then
from
there
see
if
they
pull
it
in,
because
I
believe
the
kind
tide
is
always
singular
and
then
resource
is
plural.
I
recall
correctly.
B
B
Okay
yeah,
so
if
it's,
if
it's
like
statically
typed
things
that
are
on
the,
is
that
what
you're,
referring
to
like
things
that
are
that
would
generate,
I
guess,
there's
we
could
make
those
well
yeah.
I
don't
know
I
feel
like
there's
not
enough
of
them
to
warrant
like
a
big
refactor
to
generate
them,
and
maybe
we
just
audit
them
and
change
it.
D
B
D
D
Yeah,
it
seems
pretty
trivial.
I
just
wanted
to
make
sure
to
do
it
right,
but
you
know,
for
especially
you
know,
considering
things
like,
hopefully
currently
run
online
in
english.
Maybe
that's
not
going
to
be
the
case.
Always
you
know,
and
everything
has
its
own
specific
way
to
make
plural
and
stuff
like
this.
B
Yeah,
that's
helpful,
yeah,
no
problem.
I
know
there's
like
there's
also
just
like
I
think
generally,
and
this
is
just
more
context.
The
whole
thing
there's
definitely
like
there's
some
people
are
a
bit
sensitive
about
the
fact
that
resources
are
plural
and
and
wish
they
weren't,
and
I
don't
know
that
they'll
ever
change
at
this
point,
but
I
know
it's
something
that
has
been
talked
about
in
kubernetes
land
for
a.
B
A
A
A
All
right
cool!
Well!
Thank
you
all
for
coming
to
this
community
meeting
and
I'll
see
y'all
next
week.